A Color Image Quality Evaluation Method Based On Uniform Color Coordinate System

When the human society steps into the information age, the main source of obtaining information include vision, auditory, touch, smell and taste. According to the statistic data of related scholars, it is known that the visual information consists about 70% of all information sources, and the major source of visual information is visual image. In modern society, image has been widely used in many fields, such as digital television, video conference and visual telephone and some other digital image processing systems, image has became parts of our daily lives. With the rapid development of television technology, telecommunications and computer technology, digital imaging systems standards have been improved, the "bottleneck" effect of image quality evaluation is gradually revealed. The criteria of image quality evaluation include two categories: subjective evaluation criteria and objective evaluation criteria. The subjective evaluation method means to evaluate visual effect of images by convening a group of people to grade some specifying images and dealing with those dates by statistical approach; the objective evaluation method means to confirm the physical damage of image through a series of instrument measurement parameters. However, subjective evaluation method can not be used in real-time processing, while objective evaluation method can not evaluate the images intuitively. As the final image lie on the satisfaction degree of lookers, the internationally recognized standard is subjective evaluation method (MOS: Mean Opinion Score), thus the development of image processing technology and its application are restricted.Currently, combining the subjective and objective evaluation methods of to research image quality has become a new development direction of this field, and some early scholars have made great advancements in this field. Generally speaking, the image quality evaluation related to visual images has become a new active research direction. Based on the published papers and the model and test results published by video quality experts group, image quality evaluation methods can be divided into three types: the method based on visual perception, based on the visual interest and based on the structure. During the process of analyzing of the three methods, we noted that: the literature shows that the color image quality evaluation methods associated with visual images are involved infrequently, only limit to analyze the color difference signal by the same way as the luminance signal processing methods, while no direct response from the perspective of hue and saturation. It is not conducive to acquire color image and discuss with the method of matching and visual characteristics. Therefore, we have designed a color image quality evaluation method based on uniform color coordinate system. The following areas of study are included in this paper:1. The traditional image quality assessment method includes the subjective image assessment method and the objective image assessment method. But the modern image quality assessment method can be divided into the one based on the human visual method, the one based on visual interest method, the one based on visual structure method and the one based on visual information method. In this thesis, we have analyzed the merit and shortcoming of each method.2. After discussing the color vision principle, the common color model, the principle of the three kinds of color coordinate system and the chromaticity diagram, this thesis particularly studies the CIE1976 L*a*b*uniform color space and CIE1976 L*a*b color difference formula, and generalize it to the commonly used CIEDE2000 color difference formula.3. To definitely describe a color image in image quality assessment system, three parameters including luminance, hue and saturation are needed. In order to simply the calculation procedure, it is general to employ two color difference signals to represent chromaticity factors. But color difference signal is not as intuitive as hue and saturation. Therefore, we have established a general luminance, hue and saturation computational procedure. This method enables us to use the three parameters, including luminance, hue and saturation, to exactly describe a colored image. This method can be applied to new digital video signal designing, encoding and image quality assessment and so on.4. Presently, the designing of colored test image signal is mainly based on the definition by making analogy between the module and phase angle and hue and saturation signal. However, the module and phase angle concept certainly cannot directly reflect the spike of the chromaticity physics meaning, which limits it integration with chromaticity diagram and limits the next-step research of visual correlation test image method. According to the thesis based on the television study chromaticity diagram definition, the thesis has present a three essential factors based test image signal design method, by solving the colorimetric parameter estimation problem. This method may satisfy the requirement of television signal of different types or encoding methods5. According to the CIE2000 color difference standard, and by considering the inspection color difference vision feeling, the color difference revision parameter and its the visual feeling effect, the thesis explores the relationship between the colored light three essential factors and the XYZ color coordinate system and utilizes the transformation between the XYZ chromaticity diagram and the uniform chromaticity, establishes a design method for three essential factors based CIE2000 image test signal.6. This thesis discussed the choice condition of the color difference revision parameter in CIE DE2000 color difference formula, and deduced the relationship between CIE DE2000 color difference formula and the subjective visual through massive color difference vision feeling experiment. This provides some new basis for the establishment of objective evaluation method which is closed to subjective evaluation criteria.7. Combining the image subjective evaluation criteria, and the relationship between the color difference unit and the subjective sensation, this thesis established a series of new comparison standard levels between the color difference unit and the subjective sensation and present the detailed experimental processing.