Design And Implementation Of An Stereo Effect Assessment Method For 3D Animation

Stereoscopic 3D(S3D)animation technology is widely used in visualization presentation field,such as movies,games and education.Compared with other technologies for producing S3D content,S3D animation is the naturally suitable medium to bring vivid stereo visual experience,because producers can fully control the content during the whole producing procedure.Therefore,most current 3D animation films are made by using S3D animation technology.During the production,the assessment of stereo effect experience plays an important role.Judging whether the stereo effect is good or not is mainly based on the subjective assessment of producers in practice,and the assessment standard is greatly influenced by producers' subjective factors.Meanwhile,the assessment process is likely to be affected by the testing environment and projecting equipment.Thus,a direct and objective method for stereo visual effect assessment is of great value to reduce the workload and improve the whole efficiency of 3D animation.For these reasons,we propose an assessment method that can automatically score the visual comfort and perceived depth of animation scene frames for S3D animation pre-production,which can give adjustment feedback timely and directly.In this dissertation,we first introduce some technologies and fundamentals related to stereo effect assessment,including binocular depth perception,binocular display technology,S3D animation content capturing method and assessment model construction.Visual comfort and perceived depth are two main indicators of stereo effect.We then propose a visual comfort assessment model based on machine-learning method and a perceived depth assessment method based on geometric relationship.Our method can automatically score and analyze the visual comfort and perceived depth of the input animation scene frame.Finally,user experiment and performance test of professional and commercial S3D animation scenes demonstrate that the generated scores correspond with subjective assessment scores and the proposed method can assist animators in stereo effect adjustment efficiently.