Blind Spot Detection Using Monocular Camera Motion Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current blind spot detection systems for vehicles, relying on active sensors like radar and cameras, struggle to accurately detect moving objects in the blind spot region, leading to potential dangerous situations during lane changes, and often generate false alarms that can distract drivers.
Innovation Solution
A method and device using a monocular camera to sequence images, partition them into blocks, identify moving blocks, determine movement direction and distance, group adjacent blocks, and apply false alarm detection criteria such as size, trajectory, and statistical measures to accurately identify and filter out non-threatening objects, thereby reducing false alarms and enhancing driver safety.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If active sensors like radar, sonic, and LIDAR are used for blind spot detection, then detection capability is improved, but device complexity and cost increase
Solution Approach 1:
The patent extracts the blind spot detection function from complex active sensor systems and implements it using a simple monocular camera. By taking out the essential detection capability and separating it from unnecessary active sensing components, the system achieves reliable detection with reduced complexity and lower cost.
Solution Approach 2:
The patent replaces active mechanical sensing systems (radar, sonic, LIDAR) with a passive optical camera system. This substitution eliminates the need for active transmitters and complex signal processing hardware, while maintaining detection capability through image processing algorithms that analyze motion and depth from single-camera footage.
2Device complexity
If monocular camera is used for blind spot detection, then device complexity is reduced, but measurement precision and detection accuracy deteriorate
Solution Approach 1:
The patent segments the camera image into multiple regions and blocks, analyzing each segment independently for motion detection. This segmentation allows the system to compensate for the lack of depth information by examining local motion patterns across divided regions, thereby maintaining detection accuracy despite using a simple monocular camera.
Solution Approach 2:
The patent transitions from two-dimensional image analysis to three-dimensional spatial understanding by analyzing motion trajectories and depth cues derived from image sequences. By adding the temporal dimension through motion analysis across multiple frames, the system recovers depth information that would otherwise be unavailable in a monocular setup, improving detection precision.
3Reliability
If motion detection algorithms are applied to detect moving objects, then detection capability is improved, but false alarms increase
Solution Approach 1:
The patent applies different detection criteria and quality thresholds to different local regions of the image. By evaluating motion detection results locally across segmented regions and comparing them against region-specific characteristics, the system can distinguish between legitimate moving objects and false alarms, reducing spurious detections while maintaining sensitivity.
Solution Approach 2:
The patent implements feedback mechanisms that continuously refine detection results by comparing motion detection outputs against contextual information from the image sequence. This feedback loop allows the system to correct false alarms by re-evaluating detected objects in light of additional contextual cues, thereby improving detection reliability.
Data Source
AI summary
A method for detecting moving objects in a blind spot of vehicle is provided, comprising: taking a sequence of images of said blind spot region, partitioning each of said images of said sequence into blocks, identifying moving blocks which have moved between consecutive images of said sequence, determining a direction and a distance of said movement of said moving blocks; grouping adjacent moving blocks, for which directions within a predetermined direction interval and distances within a predetermined distance interval have been determined, and determining said moving objects based on said step of grouping. A corresponding blind spot detection device is provided as well.