Video Window Detection Using Characteristic Maps
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video window detection systems struggle to accurately identify and isolate video windows, especially when they are close or overlapping, leading to difficulties in applying image enhancements without affecting graphic backgrounds.
Innovation Solution
A video window detector system that generates characteristic maps and uses edge, realness, motion, and luma intensity values to differentiate video windows from graphic content, employing coarse and fine region detection and border verification to accurately identify and isolate video windows within a display region.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If image enhancement processing is applied to the entire display region, then the video window quality is improved, but image artifacts are introduced in the graphic background regions
Solution Approach 1:
The display region is segmented into multiple regions based on region characteristic values (edge values, black level values, realness values, motion values, luma intensity values). The characteristic map generator creates a map that identifies which regions are video windows and which are graphic backgrounds, enabling selective processing of only the video window regions while excluding the graphic background regions from enhancement processing.
Solution Approach 2:
Different regions of the display are treated with different processing qualities. The video window regions receive full image enhancement processing to improve quality, while the graphic background regions are excluded from processing to prevent artifact generation. This local differentiation is achieved through region classification using characteristic values.
2Ease of operation
If automatic video window detection is implemented, then selective image processing is enabled, but detection accuracy decreases for close or overlapping windows
Solution Approach 1:
The detection process is divided into multiple stages: coarse region detection using region characteristic values to identify candidate video windows, followed by fine detection using edge value maps and other characteristic maps to refine the window boundaries. This multi-stage segmentation approach enables accurate detection even for close or overlapping windows by first identifying candidate regions broadly and then precisely defining their boundaries.
Solution Approach 2:
The detection system operates at multiple resolution levels or dimensions. The coarse region detection provides a first-level approximation, while the fine detection using edge maps and characteristic value comparisons provides a second-level refinement. This dimensional progression from coarse to fine detection enables accurate identification of video windows even in complex scenarios with overlapping or close windows.
3Measurement precision
If multiple characteristic maps are generated for detailed analysis, then detection precision is improved, but processing time increases
Solution Approach 1:
The processing is segmented into sequential stages where characteristic maps are generated and used progressively. Region characteristic values are first computed to identify candidate video windows, then only the necessary fine-grained characteristic maps (edge values, black level values) are generated for the identified regions. This segmentation avoids the need to compute all possible characteristic maps across the entire display, reducing overall processing time while maintaining precision.
Solution Approach 2:
Instead of generating all possible characteristic maps for the entire display region, the system generates characteristic maps only for the regions identified as potential video windows through coarse analysis. This partial action approach computes fine-grained characteristics only where needed, significantly reducing processing time while maintaining detection precision for the actual video window regions.
Data Source
AI summary
A video window detector includes a region characteristic determiner to generate at least one characteristic value for at least one region of a display output; a characteristic map generator to generate an image map from the at least one characteristic value for at least one region of the display output; and a window detector to detect at least one video window dependent on the image map.