Visual Masking Model Implementation for Images & VideoChi Zhang, Yuhong Wang, Sukesh Kaithakapuzha
Department of Electrical Engineering, Stanford University
Motivation Visual Masking Model
Experimental Results
Understanding the image from the human eye’s point of view.
Computational Model of Visual Masking properties of HVS based on pyschophysical data from different research papers on the topic.
Applications of Visual Masking:• Image and Video filtering for display• Video compression• Watermarking• Encryption / Steganography
Image
Higher Importance
Lower Importance
• Just-Noticeable Difference (JND) Model: Eye tracking, spatio-temporal CSF, luminance adaptation, iter-band and intra-band contrast masking, block type classification
• Visual Attention Model: Color contrast, texture contrast, (motion suppression, skin color, face detection, etc)• Weighting Model: Foveation from highest attention points
ApplicationModified
Image
JND Model
Visual Attention Model Weighting Model
ModulationInputVideo
Masking Value
Input Edge ImageBlock Type
ClassificationTexture (B)Edge (W)Plain (G)
JND Visual MaskWith
Visual Attention(Edited)
Related Work•Modeling the Masking Effect of the Human Visual System with Visual Attention ModelAnmin Liu, Maansi Verma and Weisi Lin, Nanyang Technological University, Singapore
•Estimating Just-Noticeable Distortion for VideoYuting Jia, Weisi Lin, Senior Member, IEEE, and Ashraf A. Kassim
•Modeling Visual Attention’s Modulatory Aftereffects onVisual Sensitivity and Quality EvaluationZhongkang Lu, Senior Member, IEEE, Weisi Lin, Senior Member, IEEE,Xiaokang Yang, Senior Member, IEEE, EePing Ong, and Susu Yao.
AcknowledgmentsWe would like to thank Professor Bernd Girod and TA David Chen for their feedback and support for this project.