3D Object Detection Using HOFM to Separate Flat Features
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing automotive sensors struggle to accurately detect three-dimensional (3D) objects in various weather conditions and are limited in their ability to differentiate between 3D objects and flat features, such as shadows or textures, due to computational intensity and limitations in angle estimation and range detection.
Innovation Solution
A system utilizing semi-dense regular optical flow feature sets, including histogram of optical flow magnitudes (HOFM) and histogram of normalized optical flow gradients (HOFG), combined with learning algorithms, to identify and classify 3D objects from video streams, employing hardware and software for rapid computation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If computational algorithms are used to detect 3D objects from video streams, then object classification accuracy is improved, but computational intensity increases
Solution Approach 1:
The patent segments the video processing task into distinct stages: optical flow computation, histogram of optical flow magnitudes (HOFM) extraction, and histogram of normalized optical flow gradients (HOFG) extraction. Each stage processes specific features independently, allowing for optimized computational resources at each step while maintaining high classification accuracy for 3D object detection.
Solution Approach 2:
The patent applies different processing qualities to different regions of the video stream by computing optical flow and histogram features selectively. Regions identified as containing potential 3D objects receive more intensive processing with full HOFM and HOFG computation, while other regions use reduced processing, thereby balancing accuracy with computational load.
2Reliability
If sensors are increased in number and variety to improve detection capability, then detection reliability is improved, but device complexity increases
Solution Approach 1:
The patent makes the video camera system multi-functional by extracting multiple types of features (optical flow, HOFM, HOFG) from a single sensor input. This allows the same hardware to perform 3D object detection, flat feature detection, and motion analysis without requiring additional specialized sensors, thereby improving detection reliability while avoiding increased device complexity.
3Speed
If real-time processing is implemented for 3D object detection, then response speed is improved, but measurement precision may deteriorate
Solution Approach 1:
The patent performs preliminary computation of optical flow and histogram features (HOFM and HOFG) on video frames before final 3D object classification. These pre-computed features are stored and reused for multiple classification decisions, enabling real-time response while maintaining precision through comprehensive feature analysis that would be too computationally intensive to perform from scratch for each detection decision.
Data Source
AI summary
A method for identifying regions of interest (ROIs) includes receiving, by a processor from a video camera, a video image and computing, by the processor, an optical flow image, based on the video image. The method also includes computing, by the processor, a magnitude of optical flow image based on the video image and computing a histogram of optical flow magnitudes (HOFM) image for the video image based on the magnitude of optical flow image. Additionally, the method includes generating, by the processor, a mask indicating ROIs of the video image, based on the HOFM.