Research On Assessment Of Cartoon Images Quality

With the arriving of4G times and the rapid development of mobile terminal device, cartoon images have to communicate between different display devices of different resolutions. The retargeting algorithm is employed to adapt a source image of one resolution to be displayed in a device of a different resolution, which may introduce distortions during the retargeting process. What’s more, the images are affected by a wide variety of distortions during acquisition, compression, storage, processing, transmission, and reproduction processes, which result in perceptual quality degradation. Image quality assessment can provide a reference for image processing algorithms in order to show a better visual experience to people. Therefore, a reasonable evaluation for image quality has a very important application value. This dissertation is based on the theory of subjective and objective image assessment, focusing on the research about cartoon images subjective quality assessment methods and full-reference objective quality assessment methods. The paper’s research and achievements are as follows:1.Research on theories related to image quality assessment, establish a subjective assessment database for cartoon images. Research on popular image subjective assessment process and score processing, get the test images from distortion process and retargeting, record subjective score for these images under specific lab environment and finally establish a subjective assessment database serving for the data comparison and analysis between the objective assessment value and the subjective score.2.Research on the classic objective image quality assessment algorithms. Propose a new color image quality assessment method based on CIELAB color space. The method does an improvement on SSIM by introducing color evaluation scale and converts the quality assessment into the comparison on luminance, contrast, structure and color. Experiments show that the algorithm SSIM after adding color elements shows better consistency on human perception than simply using SSIM. 3.Propose a new cartoon images quality assessment method based on SIFT and SSIM model. For the problem that different terminal devices of different resolutions, the method use a top-down manner to establish the pixel matching association between two pictures. Experimental results show good consistency between the proposed objective metric for retargeting image and subjective assessment by human observers. It does have some significance on the optimizing of retargeting algorithm.4.Realize a prototype system for cartoon images quality assessment using Matlab GUI and its image library. The system provides a variety of algorithms to generate distortion test images for quality assessment and realize a variety of image quality assessment algorithms, analyze the subjective and objective data from different angles.