ENHANCEMENT OF MRI BRAIN IMAGES USING VARIOUS HISTOGRAM EQUALIZATION TECHNIQUES 1 S.Chokkalingam 2 M.Geethalakshmi 1 Assistant Professor, Dept. of CS, Research scholar, NPR Arts and Science Gandhigram Rural Institute 2 M.Phil College, Natham Deemed to be University, Gandhigram chokkalingamcs@gmail.com geethalakshmi449gmail.com ABSTRACT Image enhancement could be a mean because the enhancement of an image looks by increasing ascendance of some options or by decreasing ambiguity between totally different regions of the image. Several images suffered from poor contrast. It s necessary to enhance the contrast of images. Histogram equalization (HE) is widely used for contrast enhancement. However, it tends to alter the brightness of an image and thus, not appropriate for client electronic products, wherever conserving the first brightness is important to avoid annoying artifacts. The various techniques of HE specifying their contribution to enhance the contrast of digital image and their results are compared. To evaluate the effectiveness of our methodology we decide to use widely used metrics, Mean Square Error (MSE) and Peak Signal to Noise ratio (PSNR). Based on results of these two metrics this algorithm is proved as a flexible and effective way for medical image enhancement and can be used as a pre-processing step for medical image understanding and analysis. Keywords Image Enhancement, Histogram Equalization (HE), Contrast Enhancement, and MRI brain images. I. INTRODUCTION Generally medical images were poor quality in contrast and contain a lot of uncertainties. Medical imageenhancement improves visual appearance and enables identification of the desired region of the image [1]. Medical imaging plays a leading role in modern diagnosis and contrast enhancements of medical images are useful in helping the radiologist or surgeons to detect pathologic or abnormal regions. Medical image enhancement processing is superior for present plain human tissues and organs and is moreover excellent for aided diagnosis. In recent years different contrast enhancement algorithms are proposed by many researchers. There are many defects in the traditional image contrast enhancement techniques which usually processes the whole image and will hide both partial and specific information partly, and these methods will be inferred by the noise easily. Therefore, it cannot meet the requirements of the medical image processing. The conventional metrics, such as the peak signal-tonoise ratio(psnr) and the mean-squared error (MSE) S. Chokkalingam And M. Geethalakshmi 1
operate directly on theintensity of the image, and they do not correlate well with the subjectivefidelity ratings [2]. Image contrast enhancement is a classical problem inimage processing and computer vision [3]. Theenhancement is widely used for medical imageprocessing and as a preprocessing step in speechrecognition, texture synthesis, and many otherimage/video processing applications [4, 5,6,7].Existing enhancement approaches fall into two broad categories: spatial domain methods and frequencydomain methods [8]. The spatial domain techniqueshave gained more popularity as they are based on directmanipulation of pixels in an image and arestraightforward for visualizing the effect. In the last fewdecades, myriad spatial domain methods have beendeveloped for this purpose. Rest of the paper is organized as follows, Section I contain the introduction about our paper, Section II contains necessary background details needed for pursuing about report including medical images, image enhancement, contrast enhancement. Section III explained about various multispectral image enhancement methods. In Section IV, the results of those methods are discussed with explanations and we concluded our paper in SectionV. II. BACKGROUND 2.1 Magnetic Resonance Imaging (MRI) MRI has been used wide in research project and treatment. Medical brain images is a vigorous analysis topic in computer vision where magnetic resonance imaging has provided significant information regarding brain tissue at terribly high resolutions to be used in fields like reparative surgery, radiotherapy, stereotactic neuro surgery et al. 2.2 Image enhancement Image enhancement is the foremost process within the field of Digital Image Processing (DIP) [9]. It is terrible helpful in increasing the clarity, sharpness and extraction of features in an image. The foremost aim of image enhancement is to manipulate the actual image where the processed image should well suit for any specific quite applications. The image enhancement strategies are classified into spatial and frequency domain methods. A frequency domain method transforms the image into the frequency domain and operates with it. Some of the transformation methods are wavelet transform, Discrete Cosine Transforms (DCT) and Discrete Fourier Transform (DFT). The technique which has direct manipulation of the pixels is known as spatial domain techniques [10]. Spatial domain processes are defined by the following expression. g(x,y) = T[f(x,y)] (1) Where, g(x, y)is input image, f(x, y)is the processed image and the T is transformation function. In spatial domain techniques, the central pixel values are calculated with respect to their neighborhood pixels. This action is iteratively carried out on the whole image pixels and results the modified image. 2.3 Contrast enhancement Contrast enhancement is necessary for various image processing fields including satellite images and medical images [11]. To Contrast is defined as the range of brightness values occurred in an image. Contrast enhancement techniques are very useful for good visual perception and color reproduction. Contrast ratio of an image is calculated by the subsequent expression. Contrast Enhancement is incredibly abundant considerable once the subjective quality is a lot of necessary for the interpretation of an image. The main plagued by the affected by the contrast related drawback because of the luminance factor. Therefore, the contrast enhancement on brain images plays an important role and so it is useful for efficient information retrieval on such images. III. MRI BRAIN IMAGE ENHANCEMENT METHODS S. Chokkalingam And M. Geethalakshmi 2
3.1 Histogram Equalization Histogram Equalization is a non-linear contrast enhancement technique. It is well known contrast enhancement method that compiled with the sample distribution of the images [12]. In Histogram Equalization method, the Probability Density Function (PDF) is calculated for a given image using the subsequent expression.. Figure 1.Example of LHE (3) For k= 0, 1...L-1. Where L is maximum intensity value.xkis known as the histogram ofx. Based on the Probability Density Function, the cumulative density function (CDF) is defined using the below expression. (4) Then, the transformation function is defined using computed CDF and applied on the image to get the histogram equalized image. The transformation function is. f(x) = x₀ + (xl 1 x₀)c(x) (5) 3.2 Local Transformation Histogram Equalization (LHE) GHE takes the world wide information into consideration and can t adapt to local light condition. Local Histogram Equalization (LHE) performs block overlapped histogram equalization [13]. LHE defines a sub-block and retrieves its histogram information then histogram equalization is useful for the middle pixel using the CDF of that sub-block. After that, subblock is moved by one pixel and sub-block histogram equalization is continual till the end of the input image is reached. Thought LHE cannot adjust to partial light information, still it over-enhances some parts depending on its mask size. However, selection of an optimal block size that enhances all the part of the image isn t a simple task performs. Figure 1 shows the intensity pairs from 3X3 neighborhood windows. 3.3 Adaptive Histogram Equalization (AHE) Adaptive histogram equalization is a computer image processing technique used to improve contrast in images [14]. It differs from ordinary histogram equalization in the respect that the adaptive method computes several histograms, each corresponding to a distinct section of the image, and uses them to redistribute the lightness values of the image. Ordinary histogram equalization simply uses a single histogram for an entire image. It operates on small regions in the image, called tiles, rather than the entire image. Each tile's contrast is enhanced, so that the histogram of the output region approximately matches the histogram specified by the 'Distribution' parameter. The neighboring tiles are then combined using bilinear interpolation to eliminate artificially induced boundaries. The contrast, especially in homogeneous areas, can be limited to avoid amplifying any noise that might be present in the image. Algorithm for Adaptive Histogram Equalization Step 1:Start the program. Step 2:Obtain all the inputs like image, no of regions, dynamic range and clip limit Step 3: Pre-process the inputs Step 4:Process each contextual region producing gray level mapping Step 5:Interpolate gray level mapping in order to assemble the final image Step 6:Stop. S. Chokkalingam And M. Geethalakshmi 3
3.4 Contrast limited Adaptive Histogram Equalization (CLAHE) Contrast limited Adaptive Histogram Equalization (CLAHE) is a variant of Adaptive Histogram Equalization. It is a well-known method for reducing the noise amplification. It is a best method to enhance the shape of the image. The steps involved in CLAHE are, Step 1: The grid size is calculated based on maximum dimension Step 2: The grid points are identified throughout the given image. Step 3: Histogram for each point is calculated against the neighboring regions. Step 4: The histogram is clipped for the given clip level and this is used to calculate CDF. Step 5: Based on the CDF the transformation function is applied on each grid points with neighboring pixels. 3.5 Proposed Method: This method is able to enhance the image contrast while preserving the background brightness for images with well-defined background brightness. In this method the histogram is divided according to the foreground and the background levels The steps for performing this method are as follows. Step 1:Input the image Step 2:Find the histogram of the image Step 3:Separate the input image into sub-images based on background levels and nonbackgroundlevels range Step4:Each sub-image is then equalized independently, and Step 5: Combine the sub-images and then we get the final output image. 3.5.1 Evaluation Parameters: Two of the commonly used error metrics for qualitatively comparing various image compression The flowchart of our proposed method techniques are Peak Signal to Noise Ratio (PSNR) andmeansquare Error (MSE). Peak Signal to Noise Ratio: The PSNR is the widely used measure to find the quality of the reconstructed image. The following formula is used to calculate the PSNR value. (6) where, MAX is the maximum intensity pixel value of the image. Mean Square Error: The MSE is the cumulative squared error between the compressed and original image. The following formula is used to find the MSE value International. S. Chokkalingam And M. Geethalakshmi 4
Figure 2. Flowchart of proposed method. VI.RESULTS & DISCUSSION S. Chokkalingam And M. Geethalakshmi 5
Figure 1. Resultant images Figure 2. Histogram for resultant image shown in Figure.1 Table 1. Performance comparison of the proposed method Methods HE LHE AHE CLAHE Proposed Measurement MSE PSNR MSE PSNR MSE PSNR MSE PSNR MSE PSNR Image 1 34.3998 8.6828 90.2661 4.4932 1.4979 22.294 2.4158 20.217 0.1542 32.166 Image 2 27.4162 9.6683 84.858 4.7615 1.6992 21.7460 2.5109 20.050 0.0795 35.0472 Image 3 28.8633 9.4449 86.9306 4.6567 1.7083 21.7231 3.1377 19.0822 0.0826 34.878 Image 4 28.2725 9.5348 86.4234 4.6821 1.5694 22.0911 2.2629 20.5027 0.1975 31.0927 V.CONCLUSIONS 5) M.Pizer, The Medical Image Display andanalysis MRI brain medial images have its wide Group at the University of NorthCarolina: variety of its applications. In this paper, we Reminiscences and Philosophy, Computer Journal of reported about the concept of contrast IEEE Transactions onmedical Image, Vol. 22, No.1, pp. 2-10, 2003. enhancement and various kind of approaches to 6) A. Torre, M. Peinado, C. Segura, L. Perez-Cordoba, C. perform such enhancement on the MRI brain Benitez, and J. Rubio, Histogram Equalization of images. The MRI brain image was S. Chokkalingam And M. Geethalakshmi 6
taken and the Speech Representation for Robust Speech Recognition, popular contrast enhancement methods such as Computer Journal of IEEE Transaction on Speech Audio Histogram Equalization, Local Histogram Processions, Vol. 13, No. 3, pp. 355-366, 2005. 7) A. Wahab, H. Chin, and C. Tan Equalization, Adaptive Histogram Equalization, Novel Approach to Automated Fingerprint Recognition, in Proceedings andclahe methods are executed and their Vision, Image and Signal Processing, pp. 160-166, 1998. performances are evaluated. With our 8) C. Gonzalez and Woods E., Digital Image Processing, performance analysis, we can conclude that Addison-Wesley, 1992. proposed method gives better performance over 9) R.C. Gonzalez, and Woods, Digital image the other methods. It exhaustively leads to processing 2012. 10) P. Kalavathi, S. Boopathiraja, and Abinaya, enhance the hidden details through enhancing the Despeckling of ultrasound medical images using DW better results. and WP transform techniques, International Journal of Engineering and Technology (IJET), Vol. 9, No. 3, REFERENCES 2017. 11) K. Somasundaram, P. Kalavathi, Medical image 1) D. Nirmala, Medical image contrast enhancement contrast enhancement based on gamma correction, Int techniques, Journal of Chemical and Pharmaceutical J Knowledge Management e-learning. Vol. 3, No. 1, pp. Research, Vol.7, No.7, pp.1-8, 2015. 15-18, 2011. 2) L. Zhang, L. Zhang, X. Mou, and D. Zhang, FSIM: A 12) Y.T. Kim, Contrast enhancement using brightness Feature Similarity Index for ImageQuality Assessment, preserving bi-histogram equalization, IEEE IEEE TRANSACTIONS ON IMAGE PROCESSING, transactions on Consumer Electronics, Vol.43, No. 1, Vol. 20, No. 8, 2011. pp.1-8. 1997. 3) H. Kabir, A.A. Wadud, and O. Chae, Brightness 13) V.A. Kotkar and S.S. Gharde, ReiewOf Various Image Preserving Image Contrast Enhancement Using Contrast Enhancement Techniques,International Weighted Mixture of Global and Local Transformation Journal of Innovative Research in Science, Engineering Functions, The International Arab Journal of and Technology (IJIRSET), Vol. 2, No. 7, 2013. Information Technology, Vol. 7, No. 4, 2010. 14) N. Longkumer, M. Kumar and R. Saxena, Contrast 4) C. Pei, C.Zeng, and Chang H., VirtualRestoration of Enhancement Techniques using Histogram Ancient Chinese Paintings UsingColor Contrast Equalization: A Survey, International Journal of Enhancement and Lacuna TextureSynthesis, Computer Current Engineering and Technology (INPRESSCO), Journal of IEEETransactions Image Processing, Vol. 13, Vol. 4, No. 3, 2014. No. 3,pp. 416-429, 2004. S. Chokkalingam And M. Geethalakshmi 7