Human Detection Accuracy via Moving Object Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing human detection techniques in videos captured by network cameras struggle to accurately identify stationary humans and require a large variety of background images to improve detection accuracy, leading to inefficiencies and erroneous detections.
Innovation Solution
An image processing apparatus that includes a video obtainer, a human detector, a moving object detector, a human candidate identifier, and a determiner. This apparatus identifies human candidate areas based on the degree of matching between human detection and moving object detection images, and determines whether these areas represent humans by comparing them to a reference image of erroneously detected objects.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If moving object detection is used to detect humans, then moving humans can be detected, but stationary humans cannot be detected
Solution Approach 1:
The patent segments the detection process into multiple independent detection methods: moving object detection and human detection. By dividing the detection task into separate modules, the system can combine results from both methods to detect both moving and stationary humans, resolving the limitation of moving object detection alone.
Solution Approach 2:
The patent merges the results from moving object detection and human detection by comparing their detection results. The comparison unit integrates information from both detection methods, allowing the system to identify humans whether they are moving or stationary, thus expanding detection coverage while maintaining accuracy.
2Measurement precision
If a large variety of background images are used in the dictionary, then human detection accuracy is improved, but system complexity increases
Solution Approach 1:
The patent introduces a comparison unit as an intermediary that compares detection results from multiple sources rather than directly comparing against a large dictionary. This intermediary layer simplifies the process by filtering and validating detections before final classification, reducing the need for an extensive background image dictionary.
Solution Approach 2:
Instead of using a complete and extensive background image dictionary, the patent employs partial action by using multiple detection methods working together. The combination of moving object detection and human detection provides sufficient accuracy without requiring exhaustive background modeling, thus reducing system complexity.
3Measurement precision
If multiple detection methods are combined, then detection accuracy is improved, but processing time increases
Solution Approach 1:
The patent performs preliminary actions by conducting moving object detection and human detection in parallel before the comparison stage. By preparing detection results from multiple methods simultaneously rather than sequentially, the system reduces the total processing time while maintaining the accuracy benefits of multiple detection approaches.
Data Source
AI summary
An image processing apparatus includes a video obtainer that obtains a video captured with a camera, a human detector that performs human detection with the obtained video, a moving object detector that performs moving object detection with the obtained video, a human candidate identifier that identifies, as an image of a human candidate area, an image of an area detected through human detection by the human detector based on a degree of matching between the image of the area detected through human detection by the human detector and an image of an area detected through moving object detection by the moving object detector, and a determiner that determines whether the identified image of the human candidate area is an image of a human based on a degree of matching between the image of the human candidate area and a reference image of an object erroneously detected as a human.