Vehicle Body Damage Inspection With Spatial-Temporal Correlation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing systems struggle to accurately identify and eliminate redundant instances of vehicle damage across multiple images captured from different angles and times, leading to overestimation of the extent of damage.
Innovation Solution
A method and system that performs multi-level redundancy validation through spatial and temporal correlations of images captured by multiple sensors at varying angles and times, identifying and confirming a single physical damage region by validating persistence across frames and sensor overlap.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple image sensors capture images from different views and time points, then the coverage and detection capability of vehicle damage is improved, but the number of redundant damage identifications increases
Solution Approach 1:
The patent merges multiple candidate damage regions that correspond to the same physical location across different images into a single unified damage region. The system performs spatial correlation to match damage regions across different views and temporal correlation to match damage regions across different time points, then combines redundant detections into one representative damage region, resolving the contradiction between comprehensive detection and redundant identification.
Solution Approach 2:
The patent creates a virtual copy of the damage region by projecting damage detected in one image to corresponding locations in other images using spatial transformation matrices. This allows the system to identify and eliminate redundant damage identifications by comparing the original damage region with its projected copies across different views and time points.
2Quantity of substance
If spatial and temporal correlation processing is performed across multiple images, then redundant damage identifications are reduced, but the computational complexity and processing time increase
Solution Approach 1:
The patent performs preliminary actions by first identifying candidate damage regions in each image before performing correlation processing. The system pre-processes images to locate potential damage areas, then only performs spatial and temporal correlation on these candidate regions rather than processing entire images, reducing computational complexity while maintaining the ability to eliminate redundant identifications.
Solution Approach 2:
The patent segments the image processing task into distinct stages: candidate damage region identification, spatial correlation, temporal correlation, and redundancy elimination. This segmentation allows the system to process only relevant portions of images and apply complex correlation algorithms selectively, reducing overall computational burden while maintaining detection accuracy.
3Reliability
If candidate damage regions are identified in multiple time-spaced images, then the persistence of actual damage can be validated, but the initial number of candidate regions including false positives increases
Solution Approach 1:
The patent implements feedback mechanisms where damage regions detected in one image serve as reference points for validating detections in other images. The system uses temporal correlation to check whether candidate damage regions persist across multiple time points, providing feedback that confirms actual damage versus transient false positives, thereby improving reliability while managing the volume of candidate regions.
Data Source
AI summary
There is provided a method of image processing for detection of damage on a vehicle, comprising: accessing time-spaced image sequences depicting a region of a vehicle, captured by image sensors positioned at different heights and/or angles relative to the vehicle, identifying candidate regions of damage in the time-spaced image sequences, performing multi-level redundancy validation by: executing spatial correlation between images captured by different images sensors at different heights and/or different angles, executing temporal correlation between consecutive images captured by each image sensor, and validating persistence of each candidate region of damage across a threshold number of consecutive frames, identifying redundancy in the candidate regions of damage corresponding to a common physical location of the vehicle denoting a single physical damage region based on the multi-level redundancy validation, and providing an indication of the common physical location of the vehicle corresponding to the single physical damage region.