Digital Neuromorphic Vision for Real-Time Occupant Tracking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional image processing systems face challenges in efficiently analyzing high frame rate video data due to overwhelming computational demands, as they process all frames without prioritizing changes, leading to limited ability in real-time object detection and tracking.
Innovation Solution
A digital Neuromorphic (NM) vision system that uses a digital retina and engine to generate sparse spike data based on changes, enabling efficient object detection, classification, and tracking by compressing data and focusing on spatio-temporal differences, incorporating CMOS technology and post-processing operations like velocity vector generation and fovea tracking.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If conventional image processing systems process all frames without prioritizing changes, then they can maintain simple processing logic, but they face overwhelming computational demands and limited real-time object detection capability
Solution Approach 1:
The patent extracts and processes only the changed regions between consecutive video frames using differential image processing. Instead of analyzing all pixels in every frame, the system computes the difference between current and previous frames to identify regions of interest, thereby reducing computational complexity while maintaining real-time detection capability
Solution Approach 2:
The patent segments the video processing task into distinct stages: frame differencing to identify changes, region of interest extraction, and focused object detection only in those regions. This segmentation allows the system to avoid processing entire frames uniformly, reducing overall computational demand while improving real-time detection performance
2Quantity of substance
If high frame rate video data is compressed by capturing differences between frames, then data processing requirements are reduced, but the system must perform complex transformations and feature extraction
Solution Approach 1:
The patent applies different processing qualities to different regions of the image data. High-resolution processing is applied only to regions where changes are detected, while unchanged regions are processed at lower resolution or skipped entirely. This local quality approach reduces overall data quantity while managing transformation complexity through selective processing
Data Source
AI summary
A system and methodologies for neuromorphic vision simulate conventional analog NM system functionality and generate digital NM image data that facilitate improved object detection, classification, and tracking so as to detect and predict movement of a vehicle occupant.