Video Motion Mapping for Infrastructure-Free Object Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing object detection solutions require heavy physical infrastructure and telecommunications resources, are dependent on object connectivity, and are costly, making them impractical for widespread application.
Innovation Solution
A method that analyzes a video stream to detect moving objects by generating a motion map from differences between successive images, segmenting this map into geometric shapes, and tracking these shapes across frames, allowing for object identification and counting without requiring extensive infrastructure or connectivity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If inductive loops are placed under the road surface for vehicle detection, then vehicle detection capability is improved, but installation cost and infrastructure complexity increase significantly
Solution Approach 1:
The patent replaces the mechanical/physical inductive loop system embedded in roads with an optical/image processing system using cameras and computer vision algorithms. This substitution eliminates the need for physical infrastructure installation while maintaining vehicle detection capability through digital image analysis and object recognition techniques.
2Measurement precision
If connected objects transmit information for detection, then detection accuracy is improved, but dependency on telecommunications resources and object connectivity increases
Solution Approach 1:
The patent enables objects to be detected through their visual appearance and movement patterns captured by imaging systems, without requiring the objects themselves to transmit any information. The detection system processes visual data independently, making the system self-sufficient and eliminating dependency on object connectivity or telecommunications resources.
3Ease of manufacture
If video stream analysis is used for object detection, then infrastructure cost is reduced, but processing complexity and computational requirements increase
Solution Approach 1:
The patent segments the video stream into individual frames and further divides the image processing into distinct computational stages: background subtraction to identify moving regions, contour detection to extract object boundaries, and feature analysis to recognize objects. This segmentation of the processing pipeline reduces computational complexity by handling tasks sequentially rather than simultaneously.
Solution Approach 2:
The patent performs preliminary background modeling and subtraction before detailed object analysis. By first establishing a background model and removing static elements, the system reduces the amount of data requiring complex processing in subsequent steps, thereby lowering overall computational requirements while maintaining detection accuracy.
Data Source
Figure 1~2
Figure 3~4
Figure 5~6
AI summary
A method for detecting the movement of at least one object, electronic device, system, product, computer program, and corresponding medium. The invention relates to a method implemented in an electronic device comprising: obtaining two time-separated images capturing the same scene from a video stream; obtaining a motion map representing the differences between the components, along at least one video channel, of said images; and obtaining at least one geometric shape encompassing a set of moving portions of said motion map. The invention also relates to the electronic device, the product, computer program, and the corresponding medium.