Image Processing for Mismatch Analysis in Object Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing systems using texture-based and parallax analysis methods for object detection often fail to accurately determine the presence or absence of an object due to inconsistencies between detection results, leading to reduced performance when relying on a single detection method.
Innovation Solution
An image processing device that includes a first detection unit for texture-based detection, a second detection unit for parallax-based detection, a mismatch area extraction unit, and a mismatch area analysis unit to compare and analyze detection results, ensuring high accuracy in determining the presence or absence of objects by combining both methods.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a sensor fusion system uses a product of probability distributions from multiple detection means to determine object presence, then detection reliability is improved through comprehensive determination, but detection accuracy deteriorates when one detection method fails to detect the object
Solution Approach 1:
Instead of using the conventional approach where detection failure by one method leads to overall failure, this patent inverts the logic by treating mismatch areas (where one method detects and the other fails) as important analysis targets. The system extracts these mismatch areas and performs additional analysis to determine whether the object truly exists, thereby converting detection failures into opportunities for improved accuracy through alternative verification methods.
2Measurement precision
If a texture-based detection method with high detection performance is used, then detection accuracy is improved, but detection consistency deteriorates when the learned image does not match the current scene
Solution Approach 1:
The patent introduces a third detection method (parallax analysis) as an intermediary to mediate between texture-based detection results and ground truth. When texture-based detection achieves high accuracy but lacks consistency due to scene mismatches, the parallax analysis serves as an independent verification mechanism. By comparing results from multiple independent detection methods, the system determines object presence more reliably, reducing the impact of texture-based methods failing when learned images don't match current scenes.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The present invention provides an image processing device and an image processing method capable of improving determination accuracy of matching property with respect to two types of detections. An image processing device includes a first detection unit, a second detection unit, a mismatch area extraction unit, and a mismatch area analysis unit. The first detection unit detects a target object based on an image outputted from a sensor unit. The second detection unit detects the target object from an image outputted from the sensor unit by processing that differs from processing performed by the first detection unit. The mismatch area extraction unit compares a detection result of the first detection unit and a detection result of the second detection unit, and extracts a mismatch area where the first detection unit detects a target object and the second detection unit fails to detect the object. The mismatch area analysis unit determines the presence or absence of the target object in the mismatch area by allowing the first detection unit and the second detection unit to perform the detections different from each other.