Human Detection Using Color Uniformity for Partial Occlusion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional human detection systems using pattern recognition suffer from decreased recognition rates when a part of the human body is hidden by objects like vehicle hoods or guard rails, as the image disagreement leads to reduced robustness in detection results.
Innovation Solution
A human detection apparatus that includes a specification section, a color determination section, and a human detection section, utilizing the knowledge of uniform color of vehicle bodies and guard rails to improve recognition rates by determining the degree of color uniformity and positional relationships in the input image, thereby complementing the credibility of candidate areas.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If pattern recognition is performed using a recognition model to detect only the visible upper body part when the lower body is hidden, then the detection can be performed, but the robustness of the recognition result deteriorates because the pattern recognition is performed only for half of the estimated human image area
Solution Approach 1:
The patent introduces color uniformity as an intermediary feature to bridge the gap between visible and hidden body parts. By analyzing the color uniformity in the visible upper body region and inferring that hidden parts likely share similar color characteristics, the system compensates for missing information without directly observing those regions. This intermediary approach allows the system to maintain robust recognition despite incomplete visual information.
2Measurement precision
If the pattern recognition system requires complete human body visibility for accurate detection, then the recognition accuracy is high, but the detection fails when parts of the human body are hidden by shielding objects
Solution Approach 1:
The patent changes the evaluation parameters from requiring complete anatomical structure matching to incorporating color uniformity analysis. By shifting the recognition criteria to include color-based features that can be inferred from visible regions and applied to hidden regions, the system maintains high recognition accuracy even when parts of the body are obscured. This parameter transformation allows adaptation to partial visibility scenarios.
3Reliability
If the system performs pattern recognition on the entire estimated human image area, then complete human detection is possible, but the recognition rate deteriorates when parts are hidden by shielding objects causing disagreement with the recognition model
Solution Approach 1:
The patent applies partial action by focusing recognition efforts on the visible upper body region and using color uniformity inference to compensate for hidden areas, rather than requiring complete matching of the entire estimated human image area. This approach accepts that full model agreement cannot be achieved when parts are hidden, but uses the visible portion combined with color-based inference to maintain reliable detection.
Data Source
AI summary
The knowledge that “the color of a vehicle body or guard rail is uniform” applies to the detection of a human, improving the detection performance of the human of which a part of the body is hidden. That is, it is determined whether or not a human candidate area specified based on the ordinary human recognition model has a specific part, e.g., an area corresponding to a lower body, exhibiting the high degree of color uniformity. When affirmed, the human candidate area is understood to be “a human of which a part of the body is hidden by a hood or a trunk” and recognized as a human, same as in the case where the body is not hidden.