Mapping psycho-visual characteristics in measuring sharpness feature and blurring artifacts in video streams
AssigneeMicrosoft Corporation
InventorsWalid S. I. Ali, Rony Ferzli
Life of this patent
Estimating the term…
Abstract
A block-based, no-reference sharpness metric is provided taking advantage of Human Visual System (HVS) characteristics. Texture and smooth region blocks are excluded in computing the metric since sharpness is perceived mostly around edges. Overall sharpness metric is computed by pooling simulated combination of information in human brain employing a logistic function to replicate the behavior of HVS.
Claims19
A method to be executed at least in part in a computing device for providing a sharpness metric associated with received video quality in a video conference system, the method comprising:
- receiving a decoded video frame at a post-processing module;
- dividing the frame into blocks;
- determining edge blocks by:
- determining a number of edge pixels in each block;
- if the number of edge pixels in a block is below a predefined threshold, designating the block a smooth block;
- if the number of edge pixels in another block exceeds the predefined threshold, determining numbers of edge pixels in neighboring blocks of the other block, and
- if the numbers of edge pixels in all neighboring blocks exceed the predefined threshold, designating the other block a texture block, else
- designating the other block an edge block;
- computing local sharpness metrics for each of the edge blocks based on a kurtosis and a motion vector for each respective edge block;
- and
- computing an overall sharpness metric for the frame by pooling the local sharpness metrics over the entire frame.
A computing device capable of providing video conferencing services, the computing device comprising:
- a video capture device;
- a pre-processing module;
- an encoder;
- a decoder;
- a post-processing module configured to:
- receive a decoded video frame;
- divide the frame into blocks;
- determine smooth, texture, and edge blocks;
- exclude smooth and texture blocks;
- compute local sharpness metrics for each of the edge blocks employing an activity detection algorithm;
- and
- compute an overall sharpness metric for the frame by pooling the local sharpness metrics over the entire frame, wherein the overall sharpness metric is used to adjust at least one operational parameter of the video conference system for automatically enhance received video;
- and
- a display device for displaying the enhanced video.
A computer-readable memory device having instructions stored thereon for providing video conferencing services the instructions comprising:
- receiving a decoded video frame at a post-processing module;
- dividing the frame into blocks;
- determining edge blocks by excluding smooth and texture blocks based on a number of the edge pixels in each of the blocks and in neighboring blocks;
- computing local sharpness metrics for each of the edge blocks based on a product of a kurtosis and a motion vector for each respective edge block;
- computing an overall sharpness metric for the frame by pooling the local sharpness metrics over the entire frame and applying an exponential logistic function;
- and
- combining the overall sharpness metric with at least one from a set of:
- a blockiness metric, a jerkiness metric, and a dropped frame number to adjust at least one operational parameter of the video conference system for automatically enhancing received video quality.
Dependencies are inferred from the claim text.