Moving Image Object Detection Using Integrated Tracking Feedback
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing human detection technologies in monitoring camera videos face challenges in improving detection rates while minimizing misdetections, particularly when using methods like Histograms of Oriented Gradients for Human Detection (HOG) and object tracking systems.
Innovation Solution
An image processing apparatus and method that integrates object detection and tracking units, utilizing multiple likelihood thresholds and tracking mechanisms to enhance detection accuracy by confirming tracking targets based on object discrimination and statistical analysis of detection likelihoods.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the detection rate is increased by adjusting the discriminator threshold, then more objects are detected, but the number of misdetections increases
Solution Approach 1:
The system uses tracking results from previous frames as feedback to adjust and confirm detection results in the current frame. The tracking unit provides continuous position information that feeds back to the detection process, allowing the system to verify detected objects against expected tracking trajectories, thereby reducing misdetections while maintaining high detection rates
Solution Approach 2:
The system performs preliminary tracking in advance to establish expected object positions and trajectories before final detection confirmation. By pre-establishing tracking windows based on previous frame results, the system prepares detection candidates in advance, allowing for more accurate threshold adjustment and reduced misdetections
2Measurement precision
If detection and tracking are performed in parallel to improve apparent detection rate, then more objects are eventually detected, but the system complexity increases
Solution Approach 1:
The system merges detection and tracking functions into an integrated processing framework where detection results immediately feed into tracking, and tracking results continuously inform detection. This combining of functions allows the system to achieve high apparent detection rates through a unified process rather than separate independent systems
Solution Approach 2:
The discrimination unit serves multiple functions: it performs initial object detection, validates tracking results, and adjusts tracking parameters. This multi-functionality reduces the need for separate dedicated components for each function, thereby managing system complexity while maintaining high detection performance
Data Source
Figure 1
Figure 2
Figure 3
AI summary
An image processing apparatus includes obtaining means (100) configured to obtain a moving image, detection means (300) configured to detect a detection target object from an interest image frame in the moving image, tracking means (400) configured to obtain a tracked object in a neighborhood region of a detection position of the detection target object in an image frame preceding the interest image frame in the moving image, and determination means (700) configured to determine whether or not the tracked object corresponds to the detection target object by integrating a position where the detection target object is detected by the detection means and a position where the tracked object is obtained by the tracking means.