Video Edge Detection Using SAD and ISAD Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing edge detection methods for de-interlaced video frames, such as the 'bob' method, often incorrectly interpolate diagonal or round edges due to lack of edge information preservation, leading to poor picture quality and sensitivity to noisy data, which results in false edges and missed edges.
Innovation Solution
A method involving the calculation of sum of absolute differences (SAD) and inverse SAD (ISAD) values for selected edges in a video field to accurately detect valid edges with low computational complexity, accounting for various edge cases.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of energy
If vertical interpolation is used in the bob method, then bandwidth requirements are reduced, but edge information is lost causing stair-step artifacts
Solution Approach 1:
The patent performs edge detection before interpolation by analyzing pixel intensity transitions in the interlaced field. This preliminary identification of edge locations allows the interpolation process to preserve these edges rather than creating stair-step artifacts, thus maintaining edge quality while still using vertical interpolation for bandwidth efficiency
Solution Approach 2:
The patent applies different interpolation strategies based on local image characteristics. In regions detected as containing edges, a different interpolation approach is used compared to non-edge regions, allowing edge preservation where needed while maintaining bandwidth efficiency in other areas
2Manufacturing precision
If 3×3 pixel window correlation is used to detect edges, then edge information is preserved, but sensitivity to noisy video data increases causing false edges
Solution Approach 1:
The patent uses a larger pixel window than the minimal 3×3 size, examining more pixels around the target pixel. This excessive sampling provides redundant information that helps distinguish true edges from noise through statistical analysis, reducing false edge detection while still preserving real edge information
Solution Approach 2:
The patent incorporates a validation step where detected edges are checked against multiple criteria including consistency across different pixel windows and adherence to expected edge characteristics. This feedback mechanism filters out false edges caused by noise while preserving true edges
3Measurement precision
If pixel correlation methods are used to find edges, then edge detection is performed, but calculation complexity increases
Solution Approach 1:
The patent extracts only the essential edge detection function by focusing on pixel intensity transitions in specific directions rather than performing full pixel correlation. This extraction of the core functionality reduces calculation complexity while maintaining adequate edge detection capability for de-interlacing applications
Data Source
AI summary
A method for detecting edges in a video field, comprising the steps of: selecting edges for a pixel of the video field; determining sum of absolute differences (“SAD”) values for the selected edges; determining inverse SAD (“ISAD”) values for the selected edges; and detecting one or more certain ones of the selected edges as a valid edge as a function of the determined SAD values and the determined ISAD values.