Towards a Perceptual Quality Metric for Computer-Generated Images - PowerPoint PPT Presentation

Introduction Materials & Methods Results & Discussion Conclusion Towards a Perceptual Quality Metric for Computer-Generated Images Pierre Boulenguez Boris Airieau Mohamed-Chaker Larabi Daniel Meneveaux XLIM Lab., SIC Dept., Poitiers, France Electronic Imaging 2012 1 / 12

Introduction Materials & Methods Results & Discussion Conclusion Work Context 2 / 12

Introduction Materials & Methods Results & Discussion Conclusion Work Context Computer-generated images (CGIs) 2 / 12

Introduction Materials & Methods Results & Discussion Conclusion Work Context Computer-generated images (CGIs) scientific visualisation; 2 / 12

Introduction Materials & Methods Results & Discussion Conclusion Work Context Computer-generated images (CGIs) scientific visualisation; electronic gaming; 2 / 12

Introduction Materials & Methods Results & Discussion Conclusion Work Context Computer-generated images (CGIs) scientific visualisation; electronic gaming; lighting engineering; 2 / 12

Introduction Materials & Methods Results & Discussion Conclusion Work Context Computer-generated images (CGIs) scientific visualisation; electronic gaming; lighting engineering; ... 2 / 12

Introduction Materials & Methods Results & Discussion Conclusion Work Context Computer-generated images (CGIs) scientific visualisation; electronic gaming; lighting engineering; ... Physically Based Rendering 2 / 12

Introduction Materials & Methods Results & Discussion Conclusion Work Context Computer-generated images (CGIs) scientific visualisation; electronic gaming; lighting engineering; ... Physically Based Rendering Geometrical/wave/quantum optics 2 / 12

Introduction Materials & Methods Results & Discussion Conclusion Work Context Computer-generated images (CGIs) scientific visualisation; electronic gaming; lighting engineering; ... Physically Based Rendering Geometrical/wave/quantum optics, but: 2 / 12

Introduction Materials & Methods Results & Discussion Conclusion Work Context Computer-generated images (CGIs) scientific visualisation; electronic gaming; lighting engineering; ... Physically Based Rendering Geometrical/wave/quantum optics, but: exhaustive simulation impossible; 2 / 12

Introduction Materials & Methods Results & Discussion Conclusion Work Context Computer-generated images (CGIs) scientific visualisation; electronic gaming; lighting engineering; ... Physically Based Rendering Geometrical/wave/quantum optics, but: exhaustive simulation impossible; complexity reducing assumptions ! 2 / 12

Introduction Materials & Methods Results & Discussion Conclusion Problematic Classical Assumptions 3 / 12

Introduction Materials & Methods Results & Discussion Conclusion Problematic Classical Assumptions ambiant term; 3 / 12

Introduction Materials & Methods Results & Discussion Conclusion Problematic Classical Assumptions ambiant term; diffuse/specular scattering; 3 / 12

Introduction Materials & Methods Results & Discussion Conclusion Problematic Classical Assumptions ambiant term; diffuse/specular scattering; trichromatic; 3 / 12

Introduction Materials & Methods Results & Discussion Conclusion Problematic Classical Assumptions ambiant term; diffuse/specular scattering; trichromatic; ... 3 / 12

Introduction Materials & Methods Results & Discussion Conclusion Problematic Classical Assumptions ambiant term; diffuse/specular scattering; trichromatic; ... Compromize ? 3 / 12

Introduction Materials & Methods Results & Discussion Conclusion Problematic Classical Assumptions ambiant term; diffuse/specular scattering; trichromatic; ... Compromize ? Complexity reduction ratio of a given assumption. Quality reduction 3 / 12

Introduction Materials & Methods Results & Discussion Conclusion Problematic Classical Assumptions ambiant term; diffuse/specular scattering; trichromatic; ... Compromize ? Complexity reduction ratio of a given assumption. Quality reduction Problem 3 / 12

Introduction Materials & Methods Results & Discussion Conclusion Problematic Classical Assumptions ambiant term; diffuse/specular scattering; trichromatic; ... Compromize ? Complexity reduction ratio of a given assumption. Quality reduction Problem No metric to quantify the perceptual quality of a CGI ! 3 / 12

Introduction Materials & Methods Quality Model Results & Discussion Subjective Quality Experiment Conclusion Perceptual Quality Metric 4 / 12

Introduction Materials & Methods Quality Model Results & Discussion Subjective Quality Experiment Conclusion Perceptual Quality Metric Objective function M 4 / 12

Introduction Materials & Methods Quality Model Results & Discussion Subjective Quality Experiment Conclusion Perceptual Quality Metric Objective function M Q = M ( C 1 , . . . , C n ) M relates: 4 / 12

Introduction Materials & Methods Quality Model Results & Discussion Subjective Quality Experiment Conclusion Perceptual Quality Metric Objective function M Q = M ( C 1 , . . . , C n ) M relates: the overall perceptual quality Q , 4 / 12

Introduction Materials & Methods Quality Model Results & Discussion Subjective Quality Experiment Conclusion Perceptual Quality Metric Objective function M Q = M ( C 1 , . . . , C n ) M relates: the overall perceptual quality Q , to the perceived qualities of the criteria C 1 , . . . , C n . 4 / 12

Introduction Materials & Methods Quality Model Results & Discussion Subjective Quality Experiment Conclusion Perceptual Quality Metric Objective function M Q = M ( C 1 , . . . , C n ) M relates: the overall perceptual quality Q , to the perceived qualities of the criteria C 1 , . . . , C n . Assessment of Q and C 1 , . . . , C n 4 / 12

Introduction Materials & Methods Quality Model Results & Discussion Subjective Quality Experiment Conclusion Perceptual Quality Metric Objective function M Q = M ( C 1 , . . . , C n ) M relates: the overall perceptual quality Q , to the perceived qualities of the criteria C 1 , . . . , C n . Assessment of Q and C 1 , . . . , C n psychovisual experiments; 4 / 12

Introduction Materials & Methods Quality Model Results & Discussion Subjective Quality Experiment Conclusion Perceptual Quality Metric Objective function M Q = M ( C 1 , . . . , C n ) M relates: the overall perceptual quality Q , to the perceived qualities of the criteria C 1 , . . . , C n . Assessment of Q and C 1 , . . . , C n psychovisual experiments; automated measurement (work in progress). 4 / 12

Introduction Materials & Methods Quality Model Results & Discussion Subjective Quality Experiment Conclusion Perceptual Quality Criteria Constraints on the Criteria 5 / 12

Introduction Materials & Methods Quality Model Results & Discussion Subjective Quality Experiment Conclusion Perceptual Quality Criteria Constraints on the Criteria every possible aspect of Q ; 5 / 12

Introduction Materials & Methods Quality Model Results & Discussion Subjective Quality Experiment Conclusion Perceptual Quality Criteria Constraints on the Criteria every possible aspect of Q ; intuitive; 5 / 12

Introduction Materials & Methods Quality Model Results & Discussion Subjective Quality Experiment Conclusion Perceptual Quality Criteria Constraints on the Criteria every possible aspect of Q ; intuitive; automatically measurable. 5 / 12

Introduction Materials & Methods Quality Model Results & Discussion Subjective Quality Experiment Conclusion Perceptual Quality Criteria Constraints on the Criteria every possible aspect of Q ; intuitive; automatically measurable. Five Retained Criteria 5 / 12

Introduction Materials & Methods Quality Model Results & Discussion Subjective Quality Experiment Conclusion Perceptual Quality Criteria Constraints on the Criteria every possible aspect of Q ; intuitive; automatically measurable. Five Retained Criteria 1 Color Bleeding 5 / 12

Introduction Materials & Methods Quality Model Results & Discussion Subjective Quality Experiment Conclusion Perceptual Quality Criteria Constraints on the Criteria every possible aspect of Q ; intuitive; automatically measurable. Five Retained Criteria 1 Color Bleeding 2 Shadows 5 / 12

Introduction Materials & Methods Quality Model Results & Discussion Subjective Quality Experiment Conclusion Perceptual Quality Criteria Constraints on the Criteria every possible aspect of Q ; intuitive; automatically measurable. Five Retained Criteria 1 Color Bleeding 2 Shadows 3 Noise 5 / 12

Introduction Materials & Methods Quality Model Results & Discussion Subjective Quality Experiment Conclusion Perceptual Quality Criteria Constraints on the Criteria every possible aspect of Q ; intuitive; automatically measurable. Five Retained Criteria 1 Color Bleeding 2 Shadows 3 Noise 4 Aliasing 5 / 12