Surveillance Object Tracking With Static-Dynamic Algorithm Switching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional object tracking methods in surveillance systems lack preferred computation efficiency and accuracy, particularly in distinguishing between static and dynamic objects, leading to errors in object matching and tracking.
Innovation Solution
An object tracking method that classifies objects as static or dynamic based on bounding box overlay ratios, switching between low and high loading tracking algorithms to optimize computation efficiency and accuracy, using a surveillance apparatus with an image receiver and operation processor to analyze sequential images.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional object tracking method based on pixel changes is used, then object position can be found in the image, but the method cannot recognize the type of object or the form of object, requiring other object recognition technology to be used
Solution Approach 1:
The patent combines object tracking and object recognition functions into a single integrated system. The surveillance apparatus performs both tracking (to maintain object position across frames) and recognition (to identify object type and form) simultaneously, eliminating the need for separate recognition technology while providing comprehensive object information including position, type, and form characteristics
2Loss of information
If artificial intelligence based object tracking method is used, then object information can be detected and analyzed, but the computation efficiency and accuracy in object matching and tracking process deteriorates
Solution Approach 1:
The patent segments the object tracking process into distinct functional modules: object detection module, object recognition module, object classification module, and tracking module. Each module performs a specific function with optimized computation, avoiding the excessive computational burden of unified AI-based tracking while maintaining comprehensive object information analysis and high tracking accuracy
3Loss of energy
If single tracking algorithm is used for all objects, then the system operation load increases, but if multiple tracking algorithms are used, then the system complexity increases
Solution Approach 1:
The patent implements a dynamic algorithm selection mechanism that adapts the tracking algorithm based on object characteristics. The system classifies objects into different categories (e.g., stationary, moving, fast-moving) and automatically selects the most appropriate tracking algorithm for each category, optimizing system operation load while managing complexity through rule-based selection criteria
4Measurement precision
If object classification is not performed, then the tracking accuracy for different object types deteriorates, but if classification is performed, then the computation time increases
Solution Approach 1:
The patent performs object classification as a preliminary step before tracking. By classifying objects into different types based on their characteristics (such as motion patterns, shape, or detected features) before applying tracking algorithms, the system achieves higher tracking accuracy for each object type while minimizing computation time through pre-categorization and efficient classification rules
Data Source
AI summary
An object tracking method is applied to a surveillance apparatus and includes acquiring a first image, a second image and a third image in sequence, at least detecting a first object in the first image and at least detecting a second object in the second image, deciding the first object and the second object are relative and then classify a static object or a dynamic object, at least detecting a current object in the third image, utilizing a low loading tracking algorithm to compare the static object with the current object, and deciding whether the current object is suitable for the low loading tracking algorithm according to a comparison result.