Multi-Camera Image Analysis Using Orientation-Based Detection Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current image analysis systems face a trade-off between high accuracy and high speed in object detection, with increased processing load and decreased speed when attempting to perform accurate pattern matching across multiple orientations in surveillance images.
Innovation Solution
An image analysis method that detects object states independently across multiple cameras and controls analysis based on these states, using a network to select optimal cameras for detection processing based on object orientation, thereby optimizing processing speed and accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If highly accurate pattern matching is performed using matching patterns for multiple orientations, then detection accuracy is improved, but processing load increases and processing speed decreases
Solution Approach 1:
The system segments the detection task by dividing it into two stages: a first detection unit that performs rough detection to identify candidate regions, and a second detection unit that performs accurate pattern matching only on those candidate regions. This segmentation allows the system to maintain high detection accuracy while reducing overall processing load and improving processing speed.
2Productivity
If the size of processing region is reduced by selecting only regions with movement, then processing speed is improved, but detection accuracy may deteriorate
Solution Approach 1:
The system performs preliminary rough detection to identify candidate regions before performing accurate pattern matching. This preliminary action creates a focused set of regions that are likely to contain targets, allowing subsequent accurate detection to be performed efficiently on a reduced set of regions rather than the entire image, thus maintaining both speed and accuracy.
3Productivity
If detection processing frequency is reduced by thinning out processing image frames, then processing speed is improved, but detection accuracy decreases
Solution Approach 1:
The system performs preliminary rough detection on all or more frequently sampled frames to identify candidate regions, then performs accurate pattern matching on fewer frames at reduced frequency. This approach maintains detection accuracy by ensuring candidate regions are identified from sufficient samples, while improving processing speed by reducing the frequency of computationally intensive accurate matching operations.
Data Source
AI summary
A control apparatus detects a state of an object independently of results of analysis of images captured by a plurality of cameras and controls analysis of the images captured by the plurality of cameras concurrently capturing the object whose state has been detected, according to the state of the object.