Subjective Quality Evaluation Method And Objective Quality Evaluation Method Based On Support Vector Machines Of Stereo Image

be there. Today, stereo image collection, compression, transmission and gathering technology has become a focus of research at home and abroad, and the application prospect is very wide, however, whether stereo image generation or transmission and display, a set of system and effective stereo image quality evaluation method is very important.In this paper, subjective and objective evaluation method of stereo image is studied based on image quality assessment development. Because objective evaluation is based on Subjective evaluation, a subjective evaluation scheme is very important. Based on the characteristics of the human eye stereo vision,a new subjective evaluation is proposed. In this paper, six aspects in details are described: Building stereo image library, the choice of evaluators,description of experimentcondition, training and testing evaluators, dataprocessing and analysis of experiment result. The method use five classification and is apply to the objective evaluation through cooperation with tianjin eye hospital and a large number of experiments.Through cooperation with tianjin eye hospital and a large number of experiments, the subjective evaluation respectively on all aspects of give a specific method, the three-dimensional images of the category five classification, and through this paper method to establish the subjective evaluation scheme to apply to the objective evaluation and the comfort of three-dimensional measurement methods, primarily validated the feasibility of this plan. The resultsuggests that subjective quality evaluation method isreasonable and can state stereo image quality accurately. For stereo image quality objective evaluation, the brightness, tonal, saturation, the structure similarity and parallax are extracted. 88 images using support vector machine network are chose . 48 images are used for training, the remaining 40 images are used for test. The result is in agreement with subjective evaluation results. The accuracy is 77% and can correctly reflect stereo image quality correctly.