Person Counting via Posture Pattern Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional person counting methods in video surveillance often result in redundant counting of the same individual due to differences in posture or camera distortion, leading to inaccurate counting results.
Innovation Solution
A person counting method that extracts images from video frames, divides them into subsets based on the presence of related images in preceding frames, estimates posture patterns, and uses an image library to assign IDs and track individuals, minimizing redundant counting by reusing IDs for similar images and assigning new IDs when necessary.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional person counting methods (pedestrian detection, clustering, regression) are used, then counting speed is maintained, but counting accuracy deteriorates due to redundant counting of the same person
Solution Approach 1:
The patent performs preliminary actions by predicting the position of persons in the current frame based on positions in previous frames, and pre-establishing an image library storing person images from historical frames. This preparation enables efficient matching and identification before actual counting occurs, reducing redundant counting while maintaining accuracy
Solution Approach 2:
The patent creates image copies by storing person images from previous frames in an image library. These copied images are then matched against detected persons in the current frame to identify and eliminate redundant counts, directly addressing the accuracy issue without sacrificing counting speed
2Measurement precision
If person images from previous frames are stored and matched, then counting accuracy is improved, but system complexity increases
Solution Approach 1:
The patent segments the video sequence into discrete frames and divides persons into different subsets based on their appearance in consecutive frames. This segmentation allows the system to manage image data in manageable portions rather than processing all historical frames simultaneously, reducing overall system complexity
Solution Approach 2:
The patent implements dynamic frame selection by only storing and matching images from immediately preceding frames rather than maintaining a static library of all historical frames. This dynamic approach adapts the image library size based on current needs, balancing accuracy requirements with system complexity constraints
Data Source
AI summary
A person counting method and a person counting system are provided. The method includes extracting a group of person images to obtain a first image set; dividing the first image set into first and second subsets based on whether a related image exists in a second image set, and reusing a person ID of the related image; estimating posture patterns of images in the first subset, and storing the images in the first subset into an image library based on person IDs and the posture patterns; and selecting a target image whose similarity to an image in the second subset is highest from the image library, reusing a person ID of the target image when the similarity is greater than a threshold, and assigning a new person ID and incrementing a person counter by 1 when the similarity is not greater than the threshold.