3D Object Detection Using Optical Flow ROI Histograms
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing automotive sensors struggle to accurately detect three-dimensional (3D) objects in various weather conditions and environments, as radar sensors are inadequate for angle estimation and boundary detection, video cameras degrade in bad weather, and ultrasonic sensors have limited range and accuracy, while deep learning algorithms and feature-based methods fail to identify objects of all shapes and sizes.
Innovation Solution
A system utilizing semi-dense regular optical flow feature sets computes histograms of optical flow magnitudes (HOFM) and normalized optical flow gradients (HOFG) to differentiate 3D objects from flat features, employing hardware and software for low-latency decision making, and uses learning algorithms to classify objects.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Volume of moving object
If radar sensors are used for detecting 3D objects, then detection range is improved, but angle estimation and boundary detection accuracy deteriorates
Solution Approach 1:
The patent combines radar sensor data with video camera data to create a hybrid detection system. The radar provides detection range and motion information, while the video camera provides angle estimation and boundary detection. By merging these complementary data sources, the system achieves both long detection range and high measurement precision for angle and boundaries.
Solution Approach 2:
The patent introduces an intermediary processing system that fuses radar and video camera data. This intermediary layer processes and integrates information from both sensors, allowing the system to leverage the strengths of each sensor type while compensating for their individual weaknesses in angle estimation and boundary detection.
2Measurement precision
If video cameras are used for detecting 3D objects, then angle estimation and boundary detection accuracy is improved, but performance deteriorates in bad weather conditions
Solution Approach 1:
The patent merges video camera data with radar sensor data to create a robust detection system. The video camera provides high-precision angle estimation and boundary detection in good conditions, while the radar provides reliable detection in bad weather. By combining these sources, the system maintains reliability across varying weather conditions while preserving measurement precision.
Solution Approach 2:
The patent dynamically adjusts the weighting and processing parameters of video camera data based on detected weather conditions. When bad weather is detected, the system reduces reliance on video camera data and increases reliance on radar data, thereby maintaining reliable performance while preserving angle estimation accuracy when conditions permit.
3Measurement precision
If ultrasonic sensors are used for detecting 3D objects, then short-range detection accuracy is improved, but detection range deteriorates
Solution Approach 1:
The patent segments the detection space into short-range and long-range zones. Ultrasonic sensors are used for short-range detection where high precision is needed, while radar sensors handle long-range detection. This segmentation allows each sensor type to operate in its optimal range, achieving both high short-range accuracy and extended detection range.
4Adaptability or versatility
If deep learning algorithms are used for object detection, then object identification capability is improved, but computational complexity and processing time increases
Solution Approach 1:
The patent applies partial deep learning processing by using traditional computer vision algorithms for initial object detection and classification, then applying deep learning only to ambiguous or critical cases. This partial application maintains good object identification capability while significantly reducing computational complexity compared to full deep learning processing of all detected objects.
5Measurement precision
If multiple sensors are deployed to improve detection accuracy, then measurement precision is improved, but system complexity and cost increases
Solution Approach 1:
The patent merges radar and video camera systems into a unified detection platform with shared processing resources. By combining these sensors and implementing data fusion algorithms, the system achieves high detection accuracy while managing complexity through integrated architecture rather than separate independent systems.
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.