Subjective evaluation of image color damage based on JPEG compression

Transcription

1 2014 Fourth International Conference on Communication Systems and Network Technologies Subjective evaluation of image color damage based on JPEG compression Xiaoqiang He Information Engineering School Communication University of China Beijing, China Abstract This paper focuses on Subjective evaluation of image color damage based on JPEG compression. In this paper we design the subjective evaluation of the visual perceptual characteristics of human eyes on color damage creatively. The main goal is to research if there is obvious difference between the color damage of different chromaticity, study the perception of human visual perceptual characteristics about the compression damage of different chromaticity of the static images, try to explore the difference in it and come up with a specific conclusion. From the results of the experiment, we finally come to the conclusion that human eyes have different sensitivity to different chromaticity, that is, more sensitive to the quality damage of the color green, but not so sensitive to the quality of the color red and blue. Keywords- image quality; color damage; JPEG compression; subjective evaluation; I. INTRODUCTION With the development of information technology and image processing, the need for information of human beings is becoming more and more diversified. Visual information attracts extensive attention with its feature that it can show the information directly an vividly. Luminance and chromaticity information are the component of visual information of image, the researches on luminance information and perceptual characteristics of human eyes on luminance become rather mature. But the researches in the field of perceptual characteristics of human eyes on color damage remain blank. At present, lossy algorithms of compression are widely used to remove visual redundancy, causing the degradation of video quality. Thus, image quality evaluation is becoming a matter of concern. There are two different categories of evaluation methods, that is, subjective evaluation and objective evaluation. Subjective evaluation method evaluate the image quality mainly according to the subjective feeling of the perceivers, it can reflect the direct feeling of human eyes correctly. As is known to all, the final receivers of images are human beings, so from this point of view, subjective evaluation method is reliable, the results come from this method is also the most accurate. The researches the previous carried out on human visual perceptual characteristics of color damage only limited to the subjective evaluation of the color image and the corresponding only-luminance image of the same content, the subjective evaluation results show that, for many samples, with the same compression damage, the subjective evaluation grade of color images is lower than gray images, thus come to the conclusion that color images can show the image quality more completely than gray images. But there is still no specific conclusion on how color information (like chromaticity) affect the image quality. This paper mainly research the visual perceptual characteristics on color damage. There are not much research achievements in this field at present, thus it requires many subjective evaluation experiments. The foundation of any objective evaluation model is to replace the complicated subjective evaluation. And one of the most important index to evaluate the objective evaluation model is that whether the conclusion this objective evaluation model comes to have much correlations with the results of visual perception of human eyes. Base on this theory, this issue conducts many subjective experiments. If the result of subjective experiments reasonable and accurate or not, can directly related to the accuracy of the visual perceptual characteristics of human eyes. So it requires this issue be reasonable and accurate in the subjective experiments, from designing, material preparation, environment evaluation, evaluation staff and the analysis of the evaluation results. The subjective evaluation on image quality rely on the subjective feelings of evaluators to evaluate the quality of the object, it can truly reveals the visual perceptual characteristics of human eyes, thus during the period of subjective evaluation designing and realization, the validity of the testing environment, testing personnel and the validity of testing materials must be ensured. The subjective evaluation experiment on image quality should be conducted in specific testing environment, show the images to be evaluated to the evaluators and record their grades, and then count the average grade of all the evaluators as the final result. II. GENERAL DISCRIPTION OF JPEG JPEG is short for Joint Photographic Experts Group, the first international image compression standard. JPEG image compression algorithms can supply good compression ability, and at the same time, has rather good reconstruction quality. It is a commonly used method of lossy compression widely used in image and video processing field. The degree of compression can be adjusted, allowing a selectable tradeoff between storage size and image quality. The JPEG compression algorithm is at its best on photographs and paintings of realistic scenes with smooth /14 $ IEEE DOI /CSNT

2 variations of tone and color. For web usage, where the amount of data used for an image is important, JPEG is very popular. JPEG/Exif is also the most common format saved by digital cameras. On the other hand, JPEG may not be as well suited for line drawings and other textual or iconic graphics, where the sharp contrasts between adjacent pixels can cause noticeable artifacts. Such images may be better saved in a lossless graphics format such as TIFF, GIF, PNG, or a raw image format. The JPEG standard actually includes a lossless coding mode, but that mode is not supported in most products. As the typical use of JPEG is a lossy compression method, which somewhat reduces the image fidelity, it should not be used in scenarios where the exact reproduction of the data is required (such as some scientific and medical imaging applications and certain technical image processing work). JPEG is also not well suited to files that will undergo multiple edits, as some image quality will usually be lost each time the image is decompressed and recompressed, particularly if the image is cropped or shifted, or if encoding parameters are changed see digital generation loss for details. To avoid this, an image that is being modified or may be modified in the future can be saved in a lossless format, with a copy exported as JPEG for distribution. JPEG uses a lossy form of compression based on the discrete cosine transform (DCT). This mathematical operation converts each frame/field of the video source from the spatial (2D) domain into the frequency domain (aka transform domain.) A perceptual model based loosely on the human psychovisual system discards high-frequency information, i.e. sharp transitions in intensity, and color hue. In the transform domain, the process of reducing information is called quantization. In simpler terms, quantization is a method for optimally reducing a large number scale (with different occurrences of each number) into a smaller one, and the transform-domain is a convenient representation of the image because the high-frequency coefficients, which contribute less to the over picture than other coefficients, are characteristically small-values with high compressibility. The quantized coefficients are then sequenced and losslessly packed into the output bitstream. Nearly all software implementations of JPEG permit user control over the compression-ratio (as well as other optional parameters), allowing the user to trade off picture-quality for smaller file size. In embedded applications (such as minidv, which uses a similar DCT-compression scheme), the parameters are preselected and fixed for the application. The compression method is usually lossy, meaning that some original image information is lost and cannot be restored, possibly affecting image quality. There is an optional lossless mode defined in the JPEG standard. However, this mode is not widely supported in products. There is also an interlaced "Progressive JPEG" format, in which data is compressed in multiple passes of progressively higher detail. This is ideal for large images that will be displayed while downloading over a slow connection, allowing a reasonable preview after receiving only a portion of the data. However, support for progressive JPEGs is not universal. When progressive JPEGs are received by programs that do not support them (such as versions of Internet Explorer before Windows 7) the software only displays the image after it has been completely downloaded. There are also many medical imaging and traffic systems that create and process 12-bit JPEG images, normally grayscale images. The 12-bit JPEG format has been part of the JPEG specification for some time, but this format is not as widely supported. III. SUBJECTIVE EXPERIMENT The main goal is to research if there is obvious difference between the color damage of different chromaticity, study the perception of human visual perceptual characteristics about the compression damage of different chromaticity of the static images, try to explore the difference in it and come up with a specific conclusion. Here is how the experiment is carried out: First, take several lossless pictures of different complexities in the format of RAW with Canon D600 SLR camera; Second, down converted the pictures into high definition size; Third, use Photoshop to change the pictures color, set the chromaticity of the pictures 0, 60, 120, 180, 240 and 300; Then, carry out JPEG compression to get damaged images with different compression rate. At last, organize subjective experiment. The experimental system is as in Figure1: Figure 1. subjective evaluation system schematic diagram All the instruments and devices used in the subjective experiment are as in table 1. No Table 1. Instruments used in the experiment Instru ments SLR camera Photo shop software JPEG compress ion software Model/version Canon D600 Version: Adobe Photoshop CS5 This version is officially distributed in 2010, the latest version is CS6 public beta, distributed in Considering functions and steadiness, we finally choose the version CS5. JPEG open source code(nonoptimized) Set compression rate according to the quality difference of human visual perception with different compression parameters and considering the different 839

3 4 High definition digital monitor image complexities, 40, 60, 70, 80. Dell 2405FPW The image given in the table above is the original picture selected at the beginning. Change the colors of the pictures given in the table above, the images we get are as in Figure 2. This subjective evaluation experimental designs one group of experiment to research the influence of human eyes subjective perception quality caused by compression damage with different chromaticity, the experimental designing details are as in table 2 Table 2. Experimental designing Description Compression parameters (compression rate) Chromaticity value Experimental images Experimental time JPEG compression and changed color images subjective evaluation experiment Red flowers 25 minutes The original pictures used in the experiment is lossless pictures in the format of RAW taken by Canon 600D SLR camera, the processing of the original pictures include: use Photoshop software to change the color the pictures, down convert the pictures, and use JPEG arithmetic compression. And the colors changed include six colors with the chromaticity value 0, 60, 120, 180, 240, 300, namely red, yellow, green, cyan, blue and pink. And the compression rates are 40, 60, 70, 80. Use improved consecutive double stimulus evaluation method, the evaluators can see the original images and damaged images repeatedly, and then grade both the original and damaged images. As the main purpose of this paper is to explore the visual perception of different color distortion under same compression damage, the improvement of the subjective evaluation method this experiment used is that, we divide the images with same compression rata and different chromaticity in one unit to be evaluated; this improvement can help the evaluators to compare the quality of images with different colors in the same compression damage more directly. This experiment invites 19 evaluators, and each of them evaluates separately. The tested images used in the subjective experiment are as in table 3. Name Red flowers Table 3. tested image and its detail description Description of the picture Red flowers with bright colors and different luminance, much detail information, many layers. Capture of the picture Figure 2.changed color images In order to get more accurate experimental results, the experimental materials will be displayed on high definition monitor, so the materials to be evaluated must be in high definition format. The pictures taken at the beginning is in the format of RAW, with the size 5184*3456, before the experiment, must down convert the pictures get from Photoshop into the format of high definition(1920*1080), and then conduct JPEG compression with the down converted pictures, use non- optimized JPEG open source code to accomplish, with their compression rates 40, 60, 70, 80, to get the compressed images in corresponding compression rates. And now all the materials need in the subjective evaluation are all ready. The subjective evaluation experiment designed in this topic is conducted according to the rules strictly set by the industry standard, this standard illustrates the image quality evaluation of cable-television and broadcast systems in detail. There are also strict rules for evaluation condition, the subjective evaluation experiment is carried out in the digital television technology research room of Communication University of China, the condition accords with the rules set by the industry standard GY/T And the materials and evaluators selected in this experiment are all representative (there are students and teachers that come from different schools), the number of the evaluators is over the minimum require of the standard. The experiment is designed to last no more than 40 minutes, in order to make sure that the experiment can be finished before visual fatigue comes. And the specific condition are as in table 4. Table 4. evaluation condition No. Item HDTV reference value 1 Watching About 3 times the 840

4 distance The peak value of the screen luminance value of screen peak value at the end of the beam value of black level peak value at the dark room value of monitor s background peak value The indoor environment light Background light and lighting lights color temperature The angel between the background light and the evaluators Distribution the evaluators of The size of the images displayed IV. height of the images 150~250cd/ 0.02 About 0.01 About 0.15 low D65 height 53 width 83 Within ±30 in the horizontal direction 24 inches EXPERIMENTAL RESULTS AND ANALYSIS According to the experimental results, we come up with the following figure 3, figure 4 and figure 5. Figure 3. the average subjective grades of all chromaticity with different compression index Figure 4. the subjective grades with different compression indexes of different chromaticity Figure 5. the subjective grades of images with the same compression rates and different chromaticity From figure 3, we can see that no matter how complicated the image is, no matter what the chromaticity is, the average subjective evaluation grades of the images fall with the compression rates rise. And figure 4 shows that this conclusion also can be used for one particular color. This conclusion we come up with also accords with the fact that compression damage gets bigger with bigger compression rate From figure 4 and figure 5, we can also find that human eyes are more sensitive to the damage of images in green color, but not so sensitive to the damage of red and blue. This shows in the line chart(figure 4) as: with the same compression damage, the subjective evaluation grades of the color red and blue are the highest of all, while the grades of green color the least, which indicates that human eyes are more sensitive to the image quality damage in green color, with the same damage, but much worse subjective perception. On the contrary, human eyes are not so sensitive to the quality damage in red and blue color, which get the highest grades with the same compression damage. V. CONCLUSION This paper takes another feature called chromaticity of image besides luminance and analyzes from image quality subjective evaluation of same content in various hues. In consequence, we discover that human eyes have different sensitivities to different hues which mean human eyes have various perceive abilities to color damage in various hues. In practical terms, color damage of three primary colors is relative 841

5 difficult to perceive for human eyes while other colors are easy to discover. Based on the results of subjective experiment and the analysis above, we can come to the conclusion that human eyes have different sensitivity to different colors, that is, more sensitive to the quality damage of green color, not so sensitive to the quality damage of the color red and blue. The research achievement would offer theory basis to improve image quality objective evaluation algorithm, and even apply to objective assess of videos. In later study, we can research the human eyes perceive ability to the other characteristic of image chromaticity called saturation in further step. REFERENCES [1] Wang Zhou Lu Liang Bovik Alan C Video quality assessment using structural distortion measurement[c]//proceedings of 2002 International Conference on Image Processing Rochester NY [2] Li Yanli Jin Donghan Jiao Bingli Several typical perceptual video quality assessment models[j] Computer Engineering and Applications (13) 66-68(in Chinese) [3] [4] Watson A B Toward a perceptual video quality metric[c]//proceedings of SPIE San Diego CA [5] GB CATI Broadcasting System Technology Criterion[S](in Chinese) (GB the technology regulations of cable telivision and broadcast syestems [S]) [6] ANDREW B W Digital images and human vision[m] [S 1 ] MIT Press [7] ATIS Video Quality Report, 2002 [8] WANG Z LU L'ALAN C B Video quality assessment based on structural distortion measurement[j] Signal processing image communication (2) [9] Wang Nannan Li Guiling Video quality evaluation models based on human visual properties[j] Journal of Image and Graphics A (6) (in Chinese) 842