Surveillance Camera Object Tracking With Dependent Feature Transformation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional image recognition technologies struggle with accurately tracking objects that change angle, become occluded, or move to similar heights, leading to misjudgment in surveillance systems.
Innovation Solution
An object classifying and tracking method that utilizes a dependent degree to transform features, using a surveillance camera with an image receiver, memory, and operation processor to analyze the dependent features of target objects across images, determining if they are the same object by computing pixel ratios, vector angles, and dependent degrees stored in a memory.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional image recognition technology analyzes certain features of target objects, then the object tracking function can be implemented, but misjudgment occurs when objects change angle, become occluded, or move to similar heights
Solution Approach 1:
The patent segments the object recognition process into multiple independent feature dimensions (classify features and dependent features). By dividing the feature analysis into orthogonal components and using a dependent degree matrix to weigh their importance, the system can accurately identify objects even when some features are obscured or changed, thereby improving reliability without proportionally increasing complexity
Solution Approach 2:
The patent transforms the object recognition approach by changing from analyzing only visual appearance features to incorporating dependent degree parameters that represent the importance relationship between different features. This parameter change allows the system to adaptively weight features based on their significance, improving identification accuracy while maintaining computational efficiency through structured parameter organization
2Measurement precision
If conventional image recognition technology tracks objects through continuous surveillance images, then the moving path can be identified, but misjudgment occurs due to object rotation, tilt, occlusion, or similar-height structures
Solution Approach 1:
The patent performs preliminary action by pre-establishing the dependent degree matrix that defines the importance relationships between different classify features and dependent features. This pre-computed structure enables the system to quickly evaluate new objects by comparing their features against the predetermined weightings, improving tracking precision without increasing the real-time detection difficulty
Solution Approach 2:
The patent introduces the dependent degree matrix as an intermediary that mediates between raw feature extraction and object identification. This intermediary structure transforms the complex feature comparison problem into a weighted evaluation process, where the dependent degree values serve as intermediaries that guide the recognition process, thereby improving measurement precision while simplifying the detection and measurement difficulty
3Measurement precision
If the surveillance system accurately identifies similar objects of different types, then object classification accuracy improves, but computation and data storage requirements increase
Solution Approach 1:
The patent extracts only the essential dependent degree relationships between feature types and stores them in a compact matrix structure. By taking out and storing only these critical weighting parameters rather than complete object feature databases, the system achieves high classification accuracy for distinguishing similar objects while minimizing data storage requirements
Solution Approach 2:
The patent changes the data representation from storing comprehensive object feature sets to storing condensed dependent degree parameters. This parameter transformation allows the system to maintain high classification precision by preserving the essential relationship information in a compressed format, thereby reducing the quantity of data that needs to be stored and processed
Data Source
AI summary
An object classifying and tracking method is applied to an image stream acquired by a surveillance camera for object identification. The object classifying and tracking method includes acquiring a first classify feature and a first dependent feature of a first target object in a first image of the image stream and a second classify feature and a second dependent feature of a second target object in a second image of the image stream, acquiring a dependent degree of the first classify feature and the second classify feature from a memory, transforming the second dependent feature via the dependent degree, and analyzing the first dependent feature and the transformed second dependent feature to determine whether the first target object and the second target object are the same object.