Pedestrian Tracking Blob Split-Merge Algorithm
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing pedestrian tracking and counting technologies using near-front top-view monitoring videos struggle to accurately count multiple pedestrians in crowded areas, resulting in low accuracy due to difficulties in distinguishing and tracking stuck individuals.
Innovation Solution
A method and device that acquire video images, compare them with background images to detect and segment blobs, split or combine them based on thresholds, and use center-of-mass coordinates for tracking and counting, employing a nearest proximity matching algorithm with a Euclidean distance cost function to improve accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional blob detection is used without splitting and combining operations, then the processing is simpler, but the counting accuracy for stuck pedestrians deteriorates
Solution Approach 1:
The patent applies segmentation by splitting large blobs into smaller candidate blobs based on width and height thresholds (TW and TH). When a blob's dimensions exceed twice the threshold, it is divided into multiple segments, allowing individual pedestrians within crowded groups to be distinguished and counted separately, thereby improving counting accuracy for stuck pedestrians.
Solution Approach 2:
The patent applies merging by combining adjacent candidate blobs that belong to the same pedestrian. By calculating center-of-mass distances and comparing them to thresholds, the system merges fragmented blob segments back into unified pedestrian representations, ensuring accurate counting while reducing over-segmentation errors.
2Measurement precision
If blob splitting is performed without subsequent combining, then individual pedestrians in crowds can be distinguished, but false segmentation of single pedestrians deteriorates
Solution Approach 1:
The system segments blobs based on dimensional thresholds to distinguish stuck pedestrians, but this may cause false segmentation of single pedestrians into multiple parts.
Solution Approach 2:
The combining operation counteracts false segmentation by merging candidate blobs that represent the same pedestrian. By checking center-of-mass distances and applying merging thresholds, the system reunites fragmented segments of individual pedestrians, ensuring that each pedestrian is counted exactly once and improving overall counting reliability.
3Productivity
If all detected blobs are counted directly, then the processing is faster, but the counting result accuracy for crowded scenes deteriorates
Solution Approach 1:
Instead of counting all detected blobs directly, the system segments large blobs into candidate blobs and then merges them according to spatial relationships. This two-stage process maintains processing efficiency while significantly improving counting accuracy for crowded scenes by properly handling stuck pedestrians through the split-merge operations.
Data Source
Figure 1~2
Figure 3
Figure 4~5
AI summary
Provided are a pedestrian tracking and counting method and device for a near-front top-view monitoring video, wherein the method includes that a video image under a current monitoring scene is acquired, the acquired video image is compared with a background image, and when it is determined that the video image is a foreground image, each blob in the foreground image is segmented and combined to acquire a target blob representing an individual pedestrian, and tracking and counting are performed according to the centre-of-mass coordinate of each target blob in a detection area to acquire the number of pedestrians under the current monitoring scene. Thus the accuracy of a counting result can be improved.