Object Tracking via Disparity Map Probability Updates
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing object tracking methods fail to accurately correct the rectangular boundary of detected objects, leading to incomplete coverage of the actual object, especially when tracking multiple same-type objects over time.
Innovation Solution
An object tracking method and device that utilize continuous disparity maps to calculate and update the boundary regions of objects by matching pixel probability maps across frames, ensuring the boundary accurately encompasses the object's outline by incorporating historic tracking data and adjusting the boundary size based on probability calculations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If object detection is performed using conventional image processing methods, then objects can be detected in the current frame, but the rectangular boundary of detected objects deviates from the actual object or becomes smaller, leading to incomplete coverage
Solution Approach 1:
The patent performs preliminary expansion of the boundary region before final object detection. By expanding the boundary region to include potential object areas that may have been missed by initial detection, the system ensures more complete coverage while maintaining detection accuracy through subsequent probability calculations.
Solution Approach 2:
The patent uses feedback from historical tracking data to correct current frame detection results. By comparing current detection with historical tracking information and updating pixel probability maps iteratively, the system refines boundary accuracy across multiple frames, resolving the contradiction between detection precision and coverage.
2Area of stationary object
If the boundary region is expanded to include the entire object, then complete coverage is achieved, but the detection precision may decrease due to inclusion of non-object pixels
Solution Approach 1:
The patent applies local quality by calculating pixel probability maps that assign different probability values to different pixels within the boundary region. Instead of treating all pixels uniformly, the system identifies high-probability pixels that belong to the object and low-probability pixels that do not, maintaining precision while ensuring coverage.
Solution Approach 2:
The patent changes parameters by updating pixel probability maps based on historical tracking data and current frame analysis. By dynamically adjusting probability values for each pixel rather than using fixed thresholding, the system adapts to object appearance changes and maintains both coverage and precision.
3Measurement precision
If tracking data from multiple frames is accumulated, then tracking accuracy improves, but the computational complexity and processing time increase
Solution Approach 1:
The patent segments the tracking problem by processing objects individually and maintaining separate pixel probability maps for each tracked object. By dividing the overall tracking task into independent object-specific subtasks, the system manages computational complexity while accumulating tracking data across multiple frames for each object.
Solution Approach 2:
The patent applies partial action by focusing computational resources on pixels within boundary regions rather than processing the entire image frame. By limiting probability calculations to relevant boundary areas and using historical data to guide current processing, the system achieves high tracking accuracy with reduced computational burden.
Data Source
AI summary
Disclosed is an object tracking method. The method includes steps of obtaining a first boundary region of a waiting-for-recognition object in the disparity map related to the current frame; calculating a probability of each valid pixel in the first boundary region so as to get a pixel probability map of the waiting-for-recognition object; obtaining historic tracking data of each tracked object, which includes identifier information of the tracked object and a pixel probability map related to each of one or more prior frame related disparity maps prior to the disparity map related to the current frame; determining identifier information of the waiting-for-recognition object, and updating the pixel probability map of the waiting-for-recognition object; and updating, based on the updated pixel probability map of the waiting-for-recognition object, the first boundary region of the waiting-for-recognition object, so as to get a second boundary region of the waiting-for-recognition object.