Automatic Method for Contrast Enhancement of Natural Color Images

Similar documents
Studying for the Enhancement Captured Images at Different Lightness Directions and Levels Distribution Based on Using Histogram Equalization Method

Mean shift segmentation on Enhanced Color image

Image Contrast Enhancement based Sub-histogram Equalization Technique without Over-equalization Noise

Image Contrast Enhancement based Sub-histogram Equalization Technique without Over-equalization Noise

CHAPTER 5 A NEAR-LOSSLESS RUN-LENGTH CODER

An Adaptive Image Denoising Method based on Thresholding

Radar emitter recognition method based on AdaBoost and decision tree Tang Xiaojing1, a, Chen Weigao1 and Zhu Weigang1 1

Image Contrast Enhancement Using Histogram Modification Technique

COMPRESSION OF TRANSMULTIPLEXED ACOUSTIC SIGNALS

Image Denoising using Fast Bilateral Filter

A New Space-Repetition Code Based on One Bit Feedback Compared to Alamouti Space-Time Code

Adaptive Fuzzy Color Interpolation

APPLICATION NOTE UNDERSTANDING EFFECTIVE BITS

Fingerprint Classification Based on Directional Image Constructed Using Wavelet Transform Domains

Data Acquisition System for Electric Vehicle s Driving Motor Test Bench Based on VC++ *

The Detection of Abrupt Changes in Fatigue Data by Using Cumulative Sum (CUSUM) Method

A New Design of Log-Periodic Dipole Array (LPDA) Antenna

Illumination Compensation for 2-D Barcode Recognition Basing Morphologic

Measurements of weld geometry using image processing technology

A study on the efficient compression algorithm of the voice/data integrated multiplexer

The Fast Haar Wavelet Transform for Signal & Image Processing

MEASUREMENT AND CONTORL OF TOTAL HARMONIC DISTORTION IN FREQUENCY RANGE 0,02-10KHZ.

Analysis of SDR GNSS Using MATLAB

A SELECTIVE POINTER FORWARDING STRATEGY FOR LOCATION TRACKING IN PERSONAL COMMUNICATION SYSTEMS

CCD Image Processing: Issues & Solutions

Multisensor transducer based on a parallel fiber optic digital-to-analog converter

Lossless image compression Using Hashing (using collision resolution) Amritpal Singh 1 and Rachna rajpoot 2

Importance Analysis of Urban Rail Transit Network Station Based on Passenger

Localized Image Segmentation and Enhancement for Meteorite Images

Fault Diagnosis in Rolling Element Bearing Using Filtered Vibration and Acoustic Signal

High Speed Area Efficient Modulo 2 1

High-Order CCII-Based Mixed-Mode Universal Filter

Design of FPGA- Based SPWM Single Phase Full-Bridge Inverter

A New Basic Unit for Cascaded Multilevel Inverters with the Capability of Reducing the Number of Switches

Selection of the basic parameters of the lens for the optic-electronic target recognition system

Design of FPGA Based SPWM Single Phase Inverter

LETTER A Novel Adaptive Channel Estimation Scheme for DS-CDMA

x 1 + x x n n = x 1 x 2 + x x n n = x 2 x 3 + x x n n = x 3 x 5 + x x n = x n

Using Color Histograms to Recognize People in Real Time Visual Surveillance

Application of Improved Genetic Algorithm to Two-side Assembly Line Balancing

Analysis of Image Compression by Minimum Relative Entropy (MRE) and Restoration through Weighted Region Growing Techniques for Medical Images

X-Bar and S-Squared Charts

SHORT-TERM TRAVEL TIME PREDICTION USING A NEURAL NETWORK

Density Slicing Reference Manual

Split and Merge Based Quantitative Approach to Select Filter Combination for Image Segmentation

History and Advancement of the Family of Log Periodic Toothed Planer Microstrip Antenna

On Parity based Divide and Conquer Recursive Functions


AC : USING ELLIPTIC INTEGRALS AND FUNCTIONS TO STUDY LARGE-AMPLITUDE OSCILLATIONS OF A PENDULUM

PROJECT #2 GENERIC ROBOT SIMULATOR

The Institute of Chartered Accountants of Sri Lanka

Testing of methods for blind estimation of noise variance on large image database

A Dual-Band Through-the-Wall Imaging Radar Receiver Using a Reconfigurable High-Pass Filter

Optimization for Hue Constant RGB Sensors

H2 Mathematics Pure Mathematics Section A Comprehensive Checklist of Concepts and Skills by Mr Wee Wen Shih. Visit: wenshih.wordpress.

Analysis and Optimization Design of Snubber Cricuit for Isolated DC-DC Converters in DC Power Grid

Outline. Motivation. Analog Functional Testing in Mixed-Signal Systems. Motivation and Background. Built-In Self-Test Architecture

Color gamut of SOCS and its comparison to Pointer's gamut

Introduction to CPM-OFDM: An Energy Efficient Multiple Access Transmission Scheme

Today s Image Capturing Needs: Going beyond Color Management

Correction of Over- and Underexposed Images Using Multiple Lighting System for Exploration Robot in Dark Environments

DIGITALLY TUNED SINUSOIDAL OSCILLATOR USING MULTIPLE- OUTPUT CURRENT OPERATIONAL AMPLIFIER FOR APPLICATIONS IN HIGH STABLE ACOUSTICAL GENERATORS

PHY-MAC dialogue with Multi-Packet Reception

Survey of Low Power Techniques for ROMs

Combined Scheme for Fast PN Code Acquisition

The Firing Dispersion of Bullet Test Sample Analysis

Performance Analysis of Channel Switching with Various Bandwidths in Cognitive Radio

WSN Node Localization Regularization Algorithm Based on Quasi Optimal Criterion Parameter Selection

Objectives. Some Basic Terms. Analog and Digital Signals. Analog-to-digital conversion. Parameters of ADC process: Related terms

Optimal Arrangement of Buoys Observable by Means of Radar

CAEN Tools for Discovery

Logarithms APPENDIX IV. 265 Appendix

Compound Controller for DC Motor Servo System Based on Inner-Loop Extended State Observer

Calibrating Car-following Model with Trajectory Data by Cell Phone

NOISE IMMUNITY RESEARCH FOR NONLINEAR DYNAMICAL SYSTEMS IDENTIFICATION BASED ON VOLTERRA MODEL IN FREQUENCY DOMAIN

Fingerprint Feature Extraction Algorithm

CFAR DETECTION IN MIMO RADARS USING FUZZY FUSION RULES IN HOMOGENEOUS BACKGROUND

A New Energy Efficient Data Gathering Approach in Wireless Sensor Networks

x y z HD(x, y) + HD(y, z) HD(x, z)

Procedia - Social and Behavioral Sciences 128 ( 2014 ) EPC-TKS 2013

Application of Image Fusion to Wireless Image Transmission

General Model :Algorithms in the Real World. Applications. Block Codes

Tehrani N Journal of Scientific and Engineering Research, 2018, 5(7):1-7

Comparison of Frequency Offset Estimation Methods for OFDM Burst Transmission in the Selective Fading Channels

Indoor Localization by Pattern Matching Method Based On Extended Database

INCREASE OF STRAIN GAGE OUTPUT VOLTAGE SIGNALS ACCURACY USING VIRTUAL INSTRUMENT WITH HARMONIC EXCITATION

Measurement of Equivalent Input Distortion AN 20

Administrivia. What we will learn

Encode Decode Sample Quantize [ ] [ ]

AkinwaJe, A.T., IbharaJu, F.T. and Arogundade, 0.1'. Department of Computer Sciences University of Agriculture, Abeokuta, Nigeria

SIDELOBE SUPPRESSION IN OFDM SYSTEMS

A Novel Small Signal Power Line Quality Measurement System

Nonlinear System Identification Based on Reduced Complexity Volterra Models Guodong Jin1,a* and Libin Lu1,b

Novel Steganography System using Lucas Sequence

4. INTERSYMBOL INTERFERENCE

Novel pseudo random number generation using variant logic framework

Methods to Reduce Arc-Flash Hazards

Financial Time Series Forecasting Using Hybrid Wavelet-Neural Model

CHAPTER 8 JOINT PAPR REDUCTION AND ICI CANCELLATION IN OFDM SYSTEMS

Subcarriers and Bits Allocation in Multiuser Orthogonal Frequency Division Multiplexing System

Transcription:

J Electr Eg Techol.2015; 10(?): 30-40 http://dx.doi.org/10.5370/jeet.2015.10.2.030 ISSN(Prit) 1975-0102 ISSN(Olie) 2093-7423 Automatic Method for Cotrast Ehacemet of Natural Color Images Shyam Lal**, A. V. Narasimhadha* ad Rahul Kumar Abstract The cotrast ehacemet is great challege i the image processig whe images are sufferig from poor cotrast problem. Therefore, i order to overcome this problem a automatic method is proposed for cotrast ehacemet of atural color images. The proposed method cosist of two stages: i first stage lightess compoet i YIQ color space is ormalized by sigmoid fuctio after the adaptive histogram equalizatio is applied o Y compoet ad i secod stage automatic color cotrast ehacemet algorithm is applied o output of the first stage. The proposed algorithm is tested o differet NASA color images, hyperspectral color images ad other types of atural color images. The performace of proposed algorithm is evaluated ad compared with the other existig cotrast ehacemet algorithms i terms of colorfuless metric ad color ehacemet factor. The higher values of colorfuless metric ad color ehacemet factor imply that the visual quality of the ehaced image is good. Simulatio results demostrate that proposed algorithm provides higher values of colorfuless metric ad color ehacemet factor as compared to other existig cotrast ehacemet algorithms. The proposed algorithm also provides better visual ehacemet results as compared with the other existig cotrast ehacemet algorithms. Keywords: Cotrast ehacemet, adaptive histogram equalizatio, sigmoid fuctio, ad lightess ehacemet. 1. Itroductio Commo ad frequetly serious disagreemets exist betwee recorded atural color images ad the direct observatio of atural scees. The huma perceptio is very excels i costructig a visual represetatio with vibrat color ad detail across the wide rage of photometric levels due to variatios i the light. I additio to the huma visio computers color are to be relatively idepedet of spectral variatios i illumiatio [1]. Whe we wat to display a color image o a display device, either the low itesity as well as medium itesity areas, which are uderexposed, or the high itesity areas, which are overexposed, caot be see by observer. I order to avoid this problem, a umber of color cotrast ehacemet techiques have bee developed durig the past decades. The color cotrast ehacemet techiques are commoly used i various applicatios where subjective quality of the image is very importat. The objective of image ehacemet is to improve visual quality of image depedig o the applicatio coditios. Cotrast is a importat factor for ay idividual estimatio of image quality. It ca be used as cotrollig tool for documetig ad presetig iformatio collectio durig examiatio. The cotrast ehacemet of image refers to the amout of color differetiatio that exists betwee various features Correspodig Author: Departmet of E&C Egg., NITK Surathkal, Idia (shyam.mtec@gmail.com) * Departmet of E&C Egg., NITK Surathkal, Idia ({dha257, rrkmalik}@ gmail.com) Received: March 30, 2014; Accepted: November 16, 2014 i digital images. It is the rage of the brightess preset i the digital image. The images havig a higher cotrast level usually display a larger degree of color scale differece as compared to lower cotrast level. The cotrast ehacemet is a process that allows image features to show up more visibly by makig best use of the color preseted o the display devices. Durig the last decade a large umber of cotrast ehacemet algorithms have bee developed for color cotrast ehacemet of images for various applicatios. These are histogram equalizatio [2], global histogram equalizatio [3], local histogram equalizatio [4], adaptive histogram equalizatio ad Cotrast Limited Adaptive histogram equalizatio [5, 6], ad other techiques ad algorithms [7-38]. Oe of the most widely used algorithms is global histogram equalizatio, the basic idea of which is to adjust the itesity histogram to approximate a uiform distributio. It treats all regios of the image equally ad, thus ofte yields poor local performace i terms of detail preservatio of image. The outlie of this paper is as follows. Sectio 2 describes literature review. Sectio 3 describes proposed method for cotrast ehacemet of atural color images. Sectio 4 gives simulatio results ad discussios to demostrate the performace of the proposed method. Fially, coclusio is draw i sectio 5. 2. Literature Review The existig cotrast ehacemet techiques for mobile commuicatio ad other real time applicatios is 30 Copyright c The Korea Istitute of Electrical Egieers This is a Ope-Access article distributed uder the terms of the Creative Commos Attributio No-Commercial Licese (http://creativecommos.org/ liceses/by-c/3.0/)which permits urestricted o-commercial use, distributio, ad reproductio i ay medium, provided the origial work is properly cited.

Shyam Lal, A.V. Narasimhadha ad Rahul Kumar fall uder two broad categories maily cotrast shapig based methods ad histogram equalizatio based methods [2]. These methods are derived from digital image processig. These methods may lead to over-ehacemet ad other artifacts such as flickerig, ad cotourig. The cotrast shapig based methods are worked o by calculatig a iput-output lumiace curve defied at every lumiace level. The shape of the curve must deped o the statistics of the image frame beig processed. For example, dark images would have a dark stretch curve applied to them. Although cotrasts shapig based methods are the most popular methods used i the cosumer electroics idustry but they caot provide a localized cotrast ehacemet which is desirable. For example, whe a dark stretch is performed, bright pixels become brighter. However, a better way to ehace darker images is to stretch ad ehace the dark regios, while leavig brighter pixels utouched [2][38]. A very popular techique for cotrast ehacemet of image is histogram equalizatio techique [2-4]. A histogram equalizatio is a techique that geerates gray map which chage the histogram of image ad redistributig all pixel values to be as close as possible to user specified desired histogram. This techique is useful for processig images that have little cotrast with equal umber of pixels to each the output gray levels. The histogram equalizatio (HE) is a method to obtai a uique iput to output cotrast trasfer fuctio based o the histogram of the iput image which results i a cotrast trasfer curve that stretches the peaks of the histogram (where more iformatio is preset) ad compresses the troughs of the histogram (where less iformatio is preset) [2]. Therefore it is a special case of cotrast shapig techique. As a stadaloe techique, histogram equalizatio is used extesively i medical imagig, satellite images ad other applicatios where the emphasis is o patter recogitio ad brigig out of hidde details. Thus histogram equalizatio results i too much ehacemet ad artifacts like cotourig which is uacceptable i cosumer electroics [5-6]. Durig the last decade a umber of techiques have bee proposed by various researchers to deal with these problems. I [8], the histogram is divided ito two parts based o the iput mea, ad each part is equalized separately. This preserves the mea value of image to a certai extet. I [9], each peak of the histogram is equalized separately. A adaptatio of HE, termed as cotrast limited adaptive histogram equalizatio (CL) [6] divides the iput image ito a umber of equal sized blocks ad the performs cotrast limited histogram equalizatio o each block. The cotrast limitig is doe by clippig the histogram before histogram equalizatio. This teds to toe dow the over ehacemet effect of histogram equalizatio ad gives a more localized ehacemet. However it is much more computatioally itesive tha histogram equalizatio. If the blocks are o-overlappig, a iterpolatio scheme is eeded to prevet blocky artifacts i the output picture. Therefore overlappig blocks ca solve this problem (every pixel is replaced by the histogram equalizatio output usig a eighborhood) but it is more computatioally itesive tha usig o-overlappig blocks. So the CL also requires a field store. Fially oe more cotrast ehacemet method that is homomorphic filter is proposed i spatial domai [2]. I this filter images ormally cosist of light reflected from objects. The basic ature of the image may be characterized by two compoets: (1) the amout of source light icidet o the scee beig viewed, ad (2) the amout of light reflected by the objects i the scee but this method does ot provide good image quality [5]. Aother method is histogram specificatio which takes a desired histogram by which the expected output image histogram ca be cotrolled [2]. However specifyig the output histogram is ot a smooth task as it varies from image to image. I [10] D. J. Jobso, Z. Rahma, ad G. A. Woodell itroduced ew algorithm to improve the brightess, cotrast ad sharpess of a image. This algorithm performs a o-liear spatial trasform that provides simultaeous dyamic rage compressio [11]. The performace of this method is compared with other existig ehacemet techiques such as histogram equalizatio ad homomorphic filterig [12]. I [13] B. V. Fut, K. Barard, M. Brockigto, ad V. Cardei have ivestigated the Multi Scale Retiex algorithm approach for image ehacemet purpose, they explored the effect of processig from theoretical stadpoit [13] ad i the same year they modified the multi-scale retiex approach to image ehacemet such that the processig is more justified from a theoretical stadpoit they suggested a ew algorithm with fewer arbitrary parameters ad prove it is more flexible [14]. I [15] A. A. Bayaty has suggested a ew method to calculate image cotrast ad evaluate image quality depedig o computig the image cotrast i edge regios. Author has itroduced robust quatitative measures to determie image quality, ad after that estimated the efficiecy of the various techiques i image processig applicatios. I [16] L. Tao, M. J. Seow ad V. K. Asari have proposed image cotrast ehacemet method to improve the visual quality of digital images that exhibit dark shadows due to the limited dyamic rages of imagig ad display devices which are icapable of hadlig high dyamic rage scees. The proposed techique processes images by applyig two separate steps: dyamic rage compressio ad local cotrast ehacemet. Dyamic rage compressio is a eighborhood depedet itesity trasformatio which is able to ehace the lumiace i dark shadows while keepig the overall toality cosistet with that of the iput image. I [17] A. J. A. Dalawy has studied the TV satellite images. These images were the same with respect to the type o the three satellites. The aalyzig these images was doe statistically by fidig the statistics distributio ad studyig the relatios betwee the mea ad the stadard deviatio of the color compoud (RGB) ad light compoet (L) for the image as http://www.jeet.or.kr 31

Automatic Method for Cotrast Ehacemet of Natural Color Images whole ad for the extracted homogeeous regios. Also author studied the cotrast of image edges depedig o sobel operator i eighbor area to the edges ad studied the cotrast as fuctio for edge fidig threshold ad foud that the Hotbird has the best results. I [18] N. Hassa ad N. Akamastu have proposed ew approach for cotrast ehacemet usig sigmoid fuctio. The objective of this ew cotrast ehacer is to scale the iput image by usig sigmoid fuctio. However this method is also have some side effects. I order to improve the performace of above metioed techique aother algorithm that is exact histogram specificatio (EHS) is proposed for cotrast ehacemet of images [19]. However this method is also have some side effects. I order to provide better result aother techique that is brightess preservig dyamic fuzzy histogram equalizatio (BPDFHE) is proposed [20]. This techique is the modificatio of the brightess preservig dyamic histogram equalizatio techique to improve its brightess preservig ad cotrast ehacemet abilities while reducig its computatioal complexity. This techique uses fuzzy statistics of digital images for their represetatio ad processig. Therefore, represetatio ad processig of images i the fuzzy domai eables the techique to hadle the iexactess of gray level values i a better way which results provide improved performace. However this techique is also havig some side effects. I [21] Celik ad Jahjadi proposed cotextual ad variatioal cotrast ehacemet for image. This algorithm ehaces the cotrast of a iput image usig iterpixel cotextual iformatio. This algorithm uses a 2-D histogram of the iput image costructed usig a mutual relatioship betwee each pixel ad its eighborig pixels. A smooth 2- D target histogram is obtaied by miimizig the sum of frobeius orms of the differeces from the iput histogram ad the uiformly distributed histogram. The ehacemet is achieved by mappig the diagoal elemets of the iput histogram to the diagoal elemets of the target histogram. This algorithm produces better ehaced images results as compared to other existig state-of-the-art algorithms but this method is also have some side effects. I [22] H. Hu ad G. Ni proposed a improved retiex algorithm for image ehacemet purpose. This algorithm has provides very good performace for specific color images i terms of color costacy, cotrast ehacemet ad computatioal cost. However, this algorithm is doe i HSV color space rather tha RGB space. Therefore a additioal step is ecessary to covert a image from RGB to HSV color space ad vice-versa. However this algorithm is also have some side effects. I [23] Y. Terai et al. proposed a retiex model for color image cotrast ehacemet purpose. I this algorithm, the lumiace sigal is processed i order to reduce the computatio time without chagig color compoets from oe format to other format. But the computatio time of this algorithm is still large due to large scale Gaussia filterig. The algorithm is oly suitable for gray images. However this algorithm is also have some side effects. I [24] L. He et al. proposed image ehacemet algorithm based o retiex theory. This algorithm replaces brightess value of each pixel by the ratio of brightess value to the average values of the eighborig pixels. The mai drawbacks of this algorithm are complexity, weakly illumiated, elarged dyamic rage. I [25] X. Dig et al. proposed a ew ehacemet algorithm for color image based o huma visual system based o adaptive filter. This algorithm utilizes color space coversio to obtai a much better visibility. This algorithm has provides better effectiveess i reducig halo ad color distortio. However, the mai drawback of this algorithm is computatioal slow algorithm. O the other had various researchers also proposed may algorithms for cotrast ehacemet i DCT based compressed domai such as alpha rootig (AR) [7], multi cotrast ehacemet (MCE) [26], modified histogram Equalizatio () [27], Adaptive Cotrast Ehacemet Based o modified Sigmoid Fuctio () [28], Multicotrast Ehacemet with Dyamic Rage Compressio () [29], Cotrast Ehacemet by Scalig (CES)[30], RGB retiex theory [31] ad other methods for cotrast ehacemet [32-35]. I order to determie image quality metric, may existig image quality assessmet algorithms use oly limited image features. Durig the past years various algorithms have bee developed by various researchers, each algorithm has its ow advatages ad disadvatages. The mai drawback of may existig image algorithms are the visual quality of the ehaced image is ot good. I order to overcome this drawback of may existig cotrast ehacemet algorithms, a ew method is proposed for cotrast ehacemet of differet types of the atural color images. The proposed algorithm provide better cotrast ehacemet results as compared to other may existig cotrast ehacemet algorithms for differet types of the atural color images such as NASA images, Hperspectral images ad other types of images. 3. Proposed Algorithm The proposed algorithm cosists of two stages: i first stage lightess compoet i YIQ color space is trasformed usig sigmoid fuctio, after the adaptive histogram equalizatio () method [6] is applied o Y compoet ad i the secod stage automatic color ehacemet algorithm is applied. The proposed algorithm is abbreviated as Automatic Color Cotrast Ehacemet (ACCE) Algorithm. The model of proposed method is show i Fig. 1. Stage-I: I the first stage the iput color image is coverted to YIQ Color space by RGB to YIQ color space 32 J Electr Eg Techol.2015;10(1): 30-40

Shyam Lal, A.V. Narasimhadha ad Rahul Kumar Iput Color Image (RGB) Color Space Trasformatio (RGB to YIQ) Cosider I ( x, = Y = 0.299R + 0.587G + 0.114B, the the ormalized Itesity is give by Illumeace Compoet [Y] Normalized by Sigmoid fuctio ad Applied Chromatic Compoets [I], [Q] I( x, I ( x, = (2) 255 After this the ormalized lightess value is trasformed by usig sigmoid fuctio is give by [35] Y p Automatic Color Cotrast Ehacemet S( x, = 1+ 1 (1 I ( x, I ( x, (3) Output Color image (RGB) Fig. 1. Block diagram of proposed algorithm trasformatio because it is used i the NSTC ad PAL televisios of differet coutries. First advatages of this format is that grayscale iformatio is separated from color data, so the same sigal ca be used for both color ad black ad white sets. I the NTSC format, image data cosists of three compoets: lumiace (Y), hue (I), ad saturatio (Q).The first compoet, lumiace, represets grayscale iformatio, while the last two compoets make up chromiace (color iformatio). Secod advatage is that it takes advatage of huma color-respose characteristics. The eye is more sesitive to chages i the orage-blue (I) rage tha i the purple-gree rage (Q), therefore less badwidth is required for Q tha for I. The color space trasformatio is give as: The plot betwee S ad I is show i Fig. 2. The processed lightess compoet is obtaied by applyig adaptive histogram equalizatio () [6] o the lightess compoet ad the processed lightess compoet is deoted as Y P for further referece. The YIQ to RGB color space trasformatio of the first stage is defied as follows R = Y p G = Y p B = Y p + 0.9563I + 0.6210Q 0.2721I 0.6474Q 1.1070I + 1.7046Q Stage - II: The automatic color cotrast ehacemet algorithm is give bellow: Step(1) Covert R G B to Y I Q Step(2) Set the followig parameters (4) Y = 0.299R + 0.587G + 0.114B I = 0.596R 0.275G 0.321B Q = 0.212R 0.523G 0.311B (1) my _ Limit = 0.5, lower _ Limit = 0.008 upper _ Limit = 0.992; my _ Limit2 = 0.04; my _ Limit3 = 0.04 After the color space trasformatio, the lumiace compoet (Y) is ormalized as follows: Step(3) Calculate Y _ Adjust, I _ Adjust, Q _ Adjust ad fid Y, I ad Q compoets 1 0.9 0.8 0.7 0.6 S 0.5 0.4 0.3 0.2 0.1 0 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 I Fig. 2. Relatioship betwee iput lightess I versus output lightess S http://www.jeet.or.kr 33

Automatic Method for Cotrast Ehacemet of Natural Color Images Y _ Adjust = my _ Limit Y I _ Adjust = my _ Limit2 I I Q _ Adjust = my _ Limit3 Q Q mea Y = Y + Y _ Adjust *(1 Y ) mea = I + I _ Adjust *(0.596 I ) mea = Q + Q _ Adjust *(0.523 Q ) Step(4) Covert image Y I Q to R G B (F 1 ), the iverse color space trasformatio from Y I Q to R G B color space is give as R G B = Y = Y = Y + 0.9563I 0.2721I 1.1070I + 0.6210Q 0.6474Q + 1.7046Q Step(5) Normalized image itesity which is give below F = F 1 *255 Step(6) Fial output image ca be calculated as F out ( F V ) max mi Vmi (5) = (6) V Where V mi = max {miimum value of R G B with impact of lower limit (0.008) } V max = max {maximum value of R G B with impact of upper limit (0.992)} 4. Simulatio Results ad Discussios I order to demostrate the performace of proposed, it is tested o differet atural color images such NASA color images, hyperspectral color images ad other types of atural color images. The proposed ad other existig algorithms are implemeted usig MATLAB software (MATLAB 7.6, release 2008a), ad 4GB RAM with I3 Processor. Two experimets are coducted o differet atural color images: i the first experimet the image quality metrics is evaluated ad i the secod experimet visual ehacemet quality of image is obtaied. I order to judge the performace of proposed the color image quality parameters such as colorfuless metric (CM), color ehacemet factor (CEF) ad CPU time are the automatic choice for the researchers. The higher values of CM ad CEF imply that the visual quality of the ehaced image is good ad a lower value of CPU time is good. The colorfuless metric (CM) ad color ehacemet factor (CEF) are defied i Eq. (7) ad Eq. (8) respectively for atural color images. These o-referece image quality metrics are used to compare the performace of proposed ad other existig cotrast ehacemet techiques such adaptive histogram equalizatio () [6], alpha rootig (AR) [7], multi cotrast ehacemet (MCE) [26], modified histogram Equalizatio () [27], Adaptive Cotrast Ehacemet Based o modified Sigmoid Fuctio () [28], Multi-cotrast Ehacemet with Dyamic Rage Compressio () [29], Cotrast Ehacemet by Scalig (CES)[30], RGB retiex theory [31]. The test atural color images used for the experimet are available o the websites http://drago.larc.asa. govt/retiex/pao/ews ad http://visio.seas.harvard. edu/hyperspec/explorei.html The colorfuless metric (CM) is o-referece image quality metric CM). It is suggested by Susstruk ad Wikler [38]. The defiitio for this metric i the color space is as give below. Let the red, gree ad blue compoets of a image be deoted by R, G ad B, respectively [30]. Cosider α=r-g ad β=(r+g)/2-b, the the colorfuless of the image is defied as 2 2 2 2 ( σ σ ) ( μ μ ) CM ( I) = α + β + 0.3* α + β (7) Where σ α ad σ β are stadard deviatios of α ad β respectively. Similarly, μ α ad μ β are their meas. The color ehacemet factor (CEF) betwee output image ad iput image is defied as: colorfu l ess( CM ) of output image CEF = (8) colorfu l ess( CM ) of iput image 4.1 Experimet 1 I this experimet the performace of proposed ACCE algorithm is tested o differet atural color images such NASA color images (image11.jpg, image22.jpg, image99. jpg, ad image110.jpg), Hyperspectral color images (image33.jpg, image44.jpg, ad image55.jpg) ad other atural color images (image66.jpg). The performace of proposed has bee evaluated ad compared with may existig cotrast ehacemet techiques such adaptive histogram equalizatio () [6], alpha rootig (AR) [7], multi cotrast ehacemet (MCE) [26], modified histogram Equalizatio () [27], Adaptive Cotrast Ehacemet Based o modified Sigmoid Fuctio () [28], Multi-cotrast Ehacemet with Dyamic Rage Compressio () [29], Cotrast Ehacemet by Scalig (CES)[30], RGB retiex theory [31] for hyperspectral color images ad other atural color images ad for NASA color images. The performace of proposed ad may other existig cotrast ehacemet techiques have bee evaluated ad 34 J Electr Eg Techol.2015;10(1): 30-40

Shyam Lal, A.V. Narasimhadha ad Rahul Kumar Table 1. Comparative performace of differet methods for Hyperspectral images & other atural image Method [6] RRT [31] MCE DRC [29] MCE [26] AR [7] [27] [28] CES [30] ACCE (Proposed) Parameters image33.jpg CM(Origial) 26.23 26.23 26.23 26.23 26.23 26.23 26.23 26.23 26.23 CM(Output) 49.35 26.77 26.22 26.23 26.57 0.72 0.45 31.87 62.16 CEF 1.88 1.02 1.00 1.00 1.01 0.03 0.02 1.21 2.37 CPU Time (secod) 0.20 28.56 4.37 0.03 0.08 0.14 0.36 0.06 0.34 image44.jpg CM(Origial) 33.49 33.49 33.49 33.49 33.49 33.49 33.49 33.49 33.49 CM(Output) 42.76 40.63 33.56 33.49 33.35 0.73 0.68 38.21 73.39 CEF 1.28 1.21 1.00 1.00 1.00 0.20 0.02 1.14 2.19 CPU Time (secod) 0.22 27.94 4.45 0.02 0.09 0.16 0.34 0.08 0.36 image55.jpg CM(Origial) 34.24 34.24 34.24 34.24 34.24 34.24 34.24 34.24 34.24 CM(Output) 33.39 48.57 34.24 34.24 34.23 0.73 0.22 43.18 82.33 CEF 0.98 1.42 1.00 1.00 1.00 0.20 0.01 1.26 2.40 CPU Time (secod) 0.20 28.08 4.24 0.02 0.08 0.14 0.34 0.06 0.34 image66.jpg CM(Origial) 33.79 33.79 33.79 33.79 33.79 33.79 33.79 33.79 33.79 CM(Output) 33.77 29.09 34.02 33.79 34.23 0.72 0.43 45.07 58.19 CEF 1.00 0.86 1.01 1.00 1.01 0.20 0.01 1.33 1.72 CPU Time (secod) 0.22 40.86 5.38 0.02 0.09 0.16 0.39 0.08 0.41 Table 2. Comparative performace of differet methods for NASA images Method [6] RRT [31] MCE DRC [29] MCE [26] AR [7] [27] [28] CES [30] ACCE (Proposed) Parameters image33.jpg CM(Origial) 51.06 51.06 51.06 51.06 51.06 51.06 51.06 51.06 51.06 CM(Output) 51.75 50.75 50.26 51.06 50.71 0.72 0.45 53.93 56.27 CEF 1.01 0.99 0.98 1.00 0.99 0.01 0.01 1.06 1.10 CPU Time (secod) 0.19 8.43 5.80 0.01 0.05 0.11 0.24 0.06 0.25 image22.jpg CM(Origial) 22.70 22.70 22.70 22.70 22.70 22.70 22.70 22.70 22.70 CM(Output) 26.16 23.49 22.34 22.70 22.22 0.73 0.38 31.27 31.82 CEF 1.15 1.03 0.98 1.00 0.98 0.03 0.02 1.38 1.40 CPU Time (secod) 0.19 8.63 2.53 0.02 0.05 0.11 0.25 0.06 0.38 image88.jpg CM(Origial) 16.58 16.58 16.58 16.58 16.58 16.58 16.58 16.58 16.58 CM(Output) 27.05 22.71 16.71 16.58 16.54 0.73 0.31 26.34 27.40 CEF 1.63 1.37 2.43 1.00 1.00 0.04 0.02 1.59 1.65 CPU Time (secod) 0.17 9.30 1.01 0.01 0.05 0.09 0.25 0.03 0.25 image99.jpg CM(Origial) 16.87 16.87 16.87 16.87 16.87 16.87 16.87 16.87 16.87 CM(Output) 22.56 19.42 17.05 16.87 16.87 0.73 0.22 21.84 30.65 CEF 1.34 1.15 1.01 1.00 1.00 0.04 0.01 1.29 1.82 CPU Time (secod) 0.20 14.16 3.48 0.01 0.08 0.13 0.28 0.03 0.31 image100.jpg CM(Origial) 17.57 17.57 17.57 17.57 17.57 17.57 17.57 17.57 17.57 CM(Output) 23.22 21.62 17.65 17.57 17.34 0.73 0.36 30.46 25.60 CEF 1.32 1.23 1.01 1.00 0.99 0.04 0.02 1.73 1.46 CPU Time (secod) 0.19 8.85 2.46 0.06 0.05 0.11 0.25 0.02 0.28 compared i terms color fulless metric (CM), color ehacemet factor (CEF) ad CPU time which are give i Table 1 ad Table 2. From Table 1 ad Table 2, it is observed that the proposed provides higher values of CM) ad (CEF) as compared to other may existig cotrast ehacemet techiques. Ad also proposed provides lower CPU time ad litter more CPU time as compared to other existig algorithm as give i Table 1 ad Table 2. 4.2 Experimet 2 This experimet visualizes subjective image ehacemet performace. The ehaced cotrast of NASA images, Hyperspectral images ad other atural color images have http://www.jeet.or.kr 35

Automatic Method for Cotrast Ehacemet of Natural Color Images (A).Origial image Rootig Fig. 3. Visual Ehacemet results of differet algorithms for image11.jpg (A).Origial image Rootig Fig. 4. Visual Ehacemet results of differet algorithms for image22.jpg (A).Origial image Rootig Fig. 5. Visual Ehacemet results of differet algorithms for image99.jpg 36 J Electr Eg Techol.2015;10(1): 30-40

Shyam Lal, A.V. Narasimhadha ad Rahul Kumar (A).Origial image Rootig Fig. 6. Visual Ehacemet results of differet algorithms for image33.jpg (A).Origial image Rootig Fig. 7. Visual Ehacemet results of differet algorithms for image44.jpg (A).Origial image Rootig Fig. 8. Visual Ehacemet results of differet algorithms for image66.jpg http://www.jeet.or.kr 37

Automatic Method for Cotrast Ehacemet of Natural Color Images bee compared with result of proposed ad may other existig cotrast ehacemet techiques. The visual cotrast ehacemet results of proposed ACCE algorithm ad may existig cotrast ehacemet techiques have bee give from Fig.3 to Fig.8. Therefore, it ca be oticed from Fig. 3(B) to Fig. 3(J), Fig. 4(B) to Fig. 4(J), Fig. 5(B) to Fig. 5(J), Fig. 6(B) to Fig. 6(J) Fig. 7(B) to Fig. 7(J) ad Fig. 8(B) to Fig. 8(J) that proposed gives better color cotrast ehacemet results as compared to other existig cotrast ehacemet techiques. 5. Coclusio I this paper, a automatic color cotrast ehacemet algorithm has bee proposed for image ehacemet purpose for various applicatios. The proposed method has bee tested o differet types of atural color images such as NASA images, Hyperspectral images ad other types of atural images. The subjective ehacemet performace of proposed has bee evaluated ad compared with other state-of-art cotrast ehacemet techiques for differet atural color images i terms of colorfuless metric (CM) ad Color ehacemet factor (CEF). The simulatio results demostrated that the proposed provides better color ehacemet quality parameters such as Colorfuless metric (CM) ad Color ehacemet factor (CEF) ad also provided better visual ehacemet results as compared to other state-of-art cotrast ehacemet techiques for differet atural color images. Therefore, the proposed performs very effectively for the cotrast ehacemet of differet atural color images. The proposed ca also be used for may other images such as remote sesig images ad eve real life photographic pictures suffer from poor cotrast as well as medium cotrast problems durig its acquisitio. Refereces [1] D. Jabso, Z. Rahma ad G. A. Woodel, A Multi- Scale Retiex for Bridgig the Gap Betwee Color Images ad the Huma Observatio of Scees, IEEE Tras. Image Processig, vol. 6, pp. 965-976, July 1997. [2] R.C. Gozalez ad R. E Woods, Digital Image Processig, Addiso-Wesley Publishig Compay, 2d editio, 1992. [3] J. A Stark, Adaptive Cotrast Ehacemet usig Geeralizatio of Histogram Equalizatio, IEEE Tras. o Image Processig, vol. 9, o. 5, pp. 889-906, 2000. [4] V. Caselles, J. L. Lisai, J. M. Morel ad G. Sapiro, Shape Preservig Local Histogram Modificatio, IEEE Tras. o Image Processig, vol. 8, o. 2, pp. 220-230, 1998. [5] S. M. Pizer, E. P. Ambur, J. D. Austi, R. Cromartie, A. Geselowitz, T. Greer, B. T. H Romey, J. B Zimmerma ad K. Zuiderveld, Adaptive Histogram Equalizatio ad its Variatios, Computer Visio, Graphics ad Image Processig, vol. 39, o. 3, pp. 355-368, 1987. [6] K. Zuiderveld, Cotrast Limited Adaptive Histogram Equalizatio, Chapter VIII.5, Graphics Gems IV, Cambridge, MA, Academic Press, pp. 474-485, 1994. [7] S. Aghagolzadeh ad O. K. Ersoy, Trasform Image Ehacemet, Optical Egieerig, vol. 31, pp. 614-626, 1992. [8] Y. T Kim, Cotrast Ehacemet usig Brightess Preservig Bi-histogram Equalizatio, IEEE Tras. o Cosumer Electroics, vol. 43, o. 1, pp. 1-8, 1997. [9] S. D. Che ad A. R. Ramli, Preservig Brightess i Histogram Equalizatio Based Cotrast Ehacemet Techiques, Digital Sigal Processig, vol. 14, o. 5, pp. 413-428, 2004. [10] Z. Rahma, G. A. Woodell, ad D. J. Jobso, A Compariso of the Multiscale Retiex with Other Image Ehacemet Techiques, I Proceedigs of the IS &T 50th Aiversary Coferece, May 1997. [11] Z. Rahma, G. A. Woodell ad D. J. Jobso, A Compariso of the Multiscale Retiex with Other Image Ehacemet Techiques, Proceedigs of the IS & T 50th Aiversary Coferece, 1997. [12] D. J. Jobso, Z. Rahma ad G. A. Woodell, Properties ad Performace of a Ceter / Surroud Retiex, IEEE Tras. o Image Processig, vol. 6, pp. 451-462, 1996. [13] B. V. Fut, K. Barard, M. Brockigto ad V. Cardei, Lumiace-Based Multi-scale Retiex, I Proceedigs of 9th Cogress of the Iteratioal Colour Associatio (AIC Color 97), Kyoto, Japa, 1997. [14] K. Barard ad B. Fut, Aalysis ad Improvemet of Multi-Scale Retiex, I Proceedigs of the IS &T/SID, Fifth Color Imagig Coferece: Color Sciece, Systems ad Applicatios, Scottsdale, Arizoa, pp. 221-226, November 17-20, 1997. [15] A. Bayaty, Adaptive Techiques for Image Co-trast Estimatio Based o Edge Detectio, Master Degree Thesis, Physics Dep., Al-Mustasiriya Uiv., 2005. [16] L. Tao, M. J. Seow ad V. K. Asari, Noliear Image Ehacemet to Improve Face, Iteratioal Joural of Computatioal Itelligece Research, vol. 2, o. 4, pp. 327-336, 2006. [17] J. A. Dalawy, A Study of TV Images Quality for Chaels Broadcast Televisio Satellite, Master of Sciece i Physics, Physics Departmet, Al- Mustasiriya Uiversity, 2008. [18] N. Hassa ad N. Akamatsu, A New Approach For Cotrast Ehacemet Usig Sigmoid Fuctio, The Iteratioal Arab Joural of Iformatio Tech-ology, vol. 1, o. 2, pp. 221-225, 2004. [19] D. Coltuc, P. Bolo ad J. M. Chasserym, Exact 38 J Electr Eg Techol.2015;10(1): 30-40

Shyam Lal, A.V. Narasimhadha ad Rahul Kumar Histogram Specificatio, IEEE Tras. o Image Processig, vol. 15, o. 5, pp. 1143-1151, 2006. [20] D. Sheet, H. Garud, A. Suveer, A. M. Mahadevappa ad J. Chatterjee, Brightess Preservig Dyamic Fuzzy Histogram Equalizatio, IEEE Tras. o Cosumer Electroics, vol. 56, o. 4, pp. 2475-2480, 2010. [21] T. Celik ad T. Tjahjad, Cotextual ad Variatioal Cotrast Ehacemet, IEEE Tras. o Image Processig, vol. 20, o. 12, pp. 3431-3441, 2011. [22] H. Hu ad G. Ni, The Improved Algorithm for the Defect of the Retiex Image Ehacemet, I Proceedigs of Iteratioal Coferece o Ati- Couterfeitig Security ad Idetificatio i Commuicatio, pp. 257-260, July, 2010. [23] Y. Terai, T. Goto, S. Hirao ad M. Sakurai, Color Image Cotrast Ehacemet by Retiex Model, I Proceedigs of IEEE 13th Iteratioal Symposium o Cosumer Electroics, pp. 392-393, May, 2009. [24] L. He, L. Luo, ad J. Shag, A Image Ehacemet Algorithm based o Retiex Theory, Proceedigs of First Iteratioal Workshop o Educatio Techology ad Computer Sciece, pp. 350-352, March, 2009. [25] X. Dig, X. Wag ad Q. Xiao, Color Image Ehacemet with a Huma Visual System based Adaptive Filter, I Proceedigs of Iteratioal Coferece o Image Aalysis ad Sigal Processig, April, 2010. [26] J. Tag, E. Peli ad S. Acto, Image Ehacemet usig A Cotrast Measure i the Compressed Domai, IEEE Sigal Processig Letter, vol. 10, o. 10, pp. 289-292, 2003. [27] H. H. Kareem, Color image with Dim regios Ehacemet Usig Modified Histogram Equalizatio Algorithm, Joural of Al-Nahrai Uiversity, vol. 15, o. 3, pp. 101-111, 2012. [28] S. Lal, ad M. Chadra, Efficiet Algorithm for Cotrast Ehacemet of Natural Images, The Iteratioal Arab Joural of Iformatio Techology, vol. 11, o. 1, pp. 96-103, 2014. [29] S. Lee, A Efficiet Cotet Based Image Ehacemet i the Compressed Domai Usig Retiex Theory, IEEE Tras. Circuits Syst. Video Techol., vol. 17, o. 2, pp. 199-213, February 2007. [30] J. Mukherjee ad S. K Mitra, Ehacemet of Color Images by Scalig the DCT Coefficiets, IEEE Tras. o Image Processig, vol. 17, o. 10, October 2008. [31] R. Kimmel, M. Elad, D. Shaked, R. Keshet, ad I. Sobel, A Variatioal Framework for Retiex, I Proceedigs of SPIE Electroic Imagig, vol. 4672, 2002. [32] M. C. Haumatharaju, M. Ravishakar, D. R. Rameshbabu ad S. Ramachadra, Color Image Ehacemet Usig Multiscale Retiex with Modified Color Restoratio Techique, I Proceedigs of Iteratioal Coferece o Emergig Applicatios of Iformatio Techology (EAIT- 2011), pp. 93-97, 2011. [33] Loza, D. Bull ad A. Achim, Automatic Cotrast Ehacemet of Low-Light Images Based o Local Statistics of Wavelet Coeciets, I Proceedigs of 17th IEEE Iteratioal Coferece o Image Processig, September 26-29, 2010. [34] S. C. Huag, F. C. Cheg, ad Y. S. Chiu, Efficiet Cotrast Ehacemet Usig Adaptive Gamma Correctio With Weightig Distributio IEEE Tras. o Image Processig, vol. 22, o. 3, pp.1032-104, March 2013. [35] E. Provezi ad V. Caselles, A Wavelet Perspective o Variatioal Perceptually-Ispired Color Ehacemet, Iteratioal Joural of Computer Visio, vol. 106, o.2, pp. 153-171, 2014. [36] Meo, K. Mehrotra, C. Moha ad S. Raka, Characterizatio of a class of Sigmoid Fuctios with applicatios to Neural Networks, Neural Networks, vol. 9, o. 5, pp. 819-8351, 1996. [37] S. J. Sagwie ad R. E. N. Hore, The Color Image Processig, Had Book, iteratioal Thomso, 1998. [38] S. Susstruk ad S. Wikler, Color Image Quality o the Iteret, I Proceedigs of IS & T/SPIE Electroic Imagig: Iteret Imagig V, vol. 5304, pp. 118-131, 2004. Shyam Lal is workig as Assistat Professor i the departmet of Electroics & Commuicatio Egieerig, Natioal Istitute of Techology Karataka, Surathkal, Magalore (KA), Idia. He has more tha 12 years of teachig ad research experiece. He has published more tha 45 papers i the area of Digital Image Processig ad Wireless Commuicatio & Computig at Iteratioal/Natioal Jourals & Cofereces. He is Member of IEEE ad other reputed professioal societies. He has bee Guest Editor of Iteratioal Joural of Sigal & Imagig System Egieerig (IJSISE), Idersciece Publishers. His area of iterest icludes Digital Image Processig, Digital Sigal Processig ad Wireless Commuicatio A.V. Narasimhadha is workig as Assistat Professor i the departmet of Electroics & Commuicatio Egieerig, Natioal Istitute of Techology Karataka, Surathkal, Magalore (KA), Idia. He has more tha 2 years of teachig ad research experiece. He has published more tha 12 papers i the area of Medical Imagig at Iteratioal/Natioal Jourals & Cofereces. He is http://www.jeet.or.kr 39

Automatic Method for Cotrast Ehacemet of Natural Color Images Member of IEEE ad other reputed professioal societies. His area of iterest icludes Medical Imagig ad Image Processig. Rahul Kumar is curretly pursuig M.Tech. i Sigal & Image Processig from, Natioal Istitute of Techology, Rourkela, Odisha, Idia. He has more tha 3 years of Teachig & Research experiece. His area of iterest icludes Digital Image Processig, Digital Sigal Processig ad Robotics. 40 J Electr Eg Techol.2015;10(1): 30-40