Image Quality Evaluation for Smart- Phone Displays at Lighting Levels of Indoor and Outdoor Conditions Optical Engineering vol. 51, No. 8, 2012 Rui Gong, Haisong Xu, Binyu Wang, and Ming Ronnier Luo Presented by Bong-Seok Choi School of Electronics Engineering Kyungpook National Univ.
Abstract Image quality assessment of mobile displays Used displays Two AMOLED and two IPS smart-phone displays Evaluation factors for image quality assessment Naturalness, colorfulness, brightness, contrast, sharpness, and overall image quality Environment to evaluate display Indoor and outdoor conditions Summary of experimental result Advancement of AMOLED display Perception of colorfulness factor because of wide color gamut Advancement of IPS display Perception of brightness, contrast, and sharpness 2/25
Introduction Image quality(iq) assessment of smart-phone displays Different of IQ perception about smart-phone displays Displayed images on smart0phone are smaller in size Including many different application type Games, GPS, e-books, etc,. 3/25
Methods to image quality assessment Physical(objective) evaluation Objective evaluation methods Mean squared error(mse), Peak signal to noise ratio(psnr), etc,. Two types of methodology in physical evaluation Full reference(fr) evaluation» Perceptible visual difference from original image No reference(nr) evaluation» Perceptible visual difference from remembered prototype Psychophysical evaluation 4/25
Aim of proposed IQ evaluation result Focus on finding critical factors when image quality assessment Finding correlation between physical parameters and perceptual IQ attributes Using 4 smart-phone displays» AMOLED and IPS panel display Characteristics of AMOLED and IPS display In-plane switching(ips) Higher peak luminance, increasing resolution and viewing angle Active matrix organic light emitting diodes(amoled) Large color gamut, thin panel thickness, and low power consumtion Evaluation factors and environment Naturalness, colorfulness, brightness, contrast, sharpness, and overall IQ Indoor and outdoor environments 5/25
Experiments Experimental equipment Using four smart-phone displays Super AMOLED, super AMOLED plus, and two IPS panel Table 1. Summary of feature of the evaluated smart-phone displays 6/25
Visualization of color gamut Standard RGB, national television standard committee, and four smart-phone displays Fig. 1. The color gamut of the four smart-phone displays in CIE1976 u v diagram 7/25
Difference between indoor and outdoor conditions Considering illumination level Most important factor affecting visual perception of displayed IQ Light condition Using typical correlated color temperature(cct) setting and two levels of ambient illumination OSRAM DULUX-L 55W/954 light module(cct of 5000K)» Indoor Illumination condition : 200lux» Outdoor illumination condition : 20000 lux 8/25
Psychophysical Procedures Composition of Experiments Observer 10 obsevers (six male and four female, age 23 to 29 years old) Experiments setting Two-minute adaptation before visual assessment Viewing distance» 25 cm Test images» Six test images each categories» Standard images and smart-phone applications 9/25
Composition of test images about six categories» Naturalness, colorfulness, brightness, contrast, sharpness, and overall IQ Fig. 2. The test image for each attribute in image quality evaluation. It contains six sub graphs, (a) Naturalness, (b) Colorfulness, (c) Brightness, (d) Contrast, (e) Sharpness, (f) Overall IQ 10/25
Composition of questionnaire Using eight-point numerical category scale Table 2. The corresponding descriptions for each numerical category 11/25
Composition of test session Repeat as number of displays» One test image on smart-phone display and evaluation of IQ» Change display with same test image Composition of whole visual experiments» 4(Phones) X 6(attribute) X 6(images) X 2(light levels) 13(10 observers + 3 repeats to test the intra-variability) 12/25
Definition of the evaluated perceptual attributes Naturalness the degree of correspondence between the visual representation of the image and a knowledge of reality as stored in memory Colorfulness Image appears to exhibit more or less chromatic color Brightness a visual sensation according to which an area appears to emit more or less light. Contrast the difference in visual properties that makes an object Distinguishable from other objects and the background. Sharpness the extent to which blurring of edges is not noticeable in an image Overall IQ the integrated perception of the overall degree of excellence of an image, including naturalness, colorfulness, brightness, contrast, and sharpness of the image 13/25
Result Observer variation Estimation of inter- and intra observer variability Using Coefficient of variation(cv) Statistical measure to represent agreement between two sets of data where n x y i i 100 2 n ( i i) CV = x y y is the number of judgments is the individual observers data is the average data for all observers 12 14/25
Mean CV values of six IQ attributes Inter- and intra-observers variability of Naturalness» Most difficult attribute for observers to scale Naturalness and contrast at 20,000 lux» Large variability with inter- and intra observers variability factor of lower CV value» Colorfulness is easier to scale at 200 and 20,000lux conditions» Reducing CV value in 20,000lx about naturalness, brightness, and sharpness» Increasing CV value in 20,000lx about colorfulness and contrast Table 2. The inter- and intra-observer variability at 200 lx and 20,000 lx 15/25
Data analysis of categorical judgment method Equal-interval scale value Adopting Case V of Thurstone`s law Converting raw data frequency matrix and its cumulative frequency matrix» Denote numbers of individual category names Calculating cumulative probability matrix and converting z-score matrix» Inverse of the standard normal cumulative distribution Calculating ultimate scale value of categorical judgment» Difference matrix and category boundary estimates of z-score value 16/25
Fig. 3. The scale values of the four smart-phone displays for each perceptual attribute at 200lx and 20,000lx. It contains six sub graphs, (a) Naturalness, (b) Colorfulness, (c) Brightness, (d) Contrast, (e) Sharpness, (f) Overall IQ 17/25
Statistical analysis of ANOVA Using two-way ANOVA analysis Analysis of displays and lighting factors Table 4. F values for significance analysis of the six image quality attributes 18/25
Image quality performances Showing opponent tendency in naturalness and colorfulness AMOLED display high performance on colorfulness and lower score on naturalness IPS display Discussions Best for naturalness and worst for colorfulness 19/25
Analysis of test result AMOLED display» Improving colorfulness arise from large color gamut» Decreasing naturalness arise from over-colorfulness IPS display» Improving naturalness and decreasing colorfulness Comparison about CCT» High CCT of AMOLED panel(9300k) and bluish white and skin color» 7000K CCT of IPS panel and suitable displayed skin color 20/25
Advantage of AMOLED with over-colorfulness» Colorfulness degrade under very high illumination condition» Advanced in smart-phone applications(game, GPS, etc,.) IPS has excellent performance on brightness and contrast» IPS panel have much better sharpness performance than AMOLED» PPI is correlative parameter for image sharpness of display 21/25
Summary of image Quality performance Beneficial parameter for IQ with smart-phone» Resolution power PPI, peak luminance, CCT and color gamut Requirement of good display at indoor and outdoor» High peak luminance, larger color gamut, high PPI value, and appropriate CCT Supplementation to evaluate IQ» Considering color appearance phenomena» Helmohltz-Kohlrausch effet (perceived brightness increases with luminance)» Hunt effect (perceived colorfulness increase with luminance) 22/25
Influence of ambient lighting levels on image quality Ambient lighting effect Significantly influencing with Brightness and contrast Important factors for smart-phone`s outdoor applications» Necessity of high peak luminance and appropriate tone reproduction characteristics for improving brightness and contrast performance of smartphone display Table 4. F values for significance analysis of the six image quality attributes 23/25
Conclusion Analysis of IQ performance about AMOLED and IPS panel Visually evaluation and analyzing two light levels IPS panel Good result of Perception of brightness, contrast and sharpness Related factors of High pixel resolution and high peak luminance AMOLED panel Achieving better performance on colorfulness» wider color gamut Degradation of naturalness» High CCT and over-colorfulness Indicating significant different result about statistical analysis of ANOVA 24/25