Object Recognition Candidate Exclusion Using Binary Image Overlap
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing object recognition techniques using template matching struggle to accurately exclude candidates in situations where multiple objects are recognized, especially when objects have different shapes or appearances, leading to erroneous exclusions or inclusions.
Innovation Solution
An object recognition processing apparatus that generates binary images based on candidate positions and attitudes, calculates the degree of overlap between candidates, and uses this overlap as a condition to exclude candidates, allowing for appropriate exclusion even with varying object shapes and appearances.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If candidate exclusion is performed based on position and attitude parameters only, then the exclusion process is simple and fast, but candidates with different shapes or appearances cannot be properly excluded leading to recognition errors
Solution Approach 1:
The patent creates binary image copies of candidates based on their position and attitude parameters, and uses these copied images for overlap comparison. This allows the system to evaluate shape-based overlap without performing complex full-image processing on all candidates, thus improving recognition accuracy while maintaining reasonable processing efficiency.
Solution Approach 2:
Instead of performing complete image processing on all candidates, the patent applies a partial action approach by first using position and attitude parameters for initial filtering, then applying binary image overlap checks only to candidates that pass the initial filter. This selective application of more complex processing improves accuracy where needed while maintaining speed for obvious cases.
2Measurement precision
If binary image generation and overlap calculation are performed for all candidates, then recognition accuracy improves by considering object shapes, but processing time increases
Solution Approach 1:
The patent segments the candidate exclusion process into two stages: first, a quick filtering stage using position and attitude parameters; second, a more accurate but time-consuming binary image overlap check applied only to candidates that pass the first stage. This segmentation reduces the number of candidates requiring intensive processing, thereby limiting time loss while maintaining high exclusion accuracy.
Solution Approach 2:
The patent applies the computationally intensive binary image generation and overlap calculation only partially - specifically, only to candidates that survive the initial position and attitude-based filtering. This partial application of the more accurate method achieves high recognition accuracy while avoiding the time penalty of applying it to all candidates.
3Adaptability or versatility
If multiple templates are used for template matching, then recognition of objects with different appearances is improved, but the number of candidates increases making exclusion more difficult
Solution Approach 1:
The patent extracts and utilizes position and attitude parameters from each candidate as key features for the exclusion process. By focusing on these extracted parameters rather than processing all candidate details equally, the system can effectively manage and exclude candidates even when the total number increases due to multiple templates, thus handling diverse object appearances without being overwhelmed by candidate quantity.
Data Source
AI summary
An object recognition processing apparatus includes: an image obtainment unit that obtains an image; a template matching unit that obtains a recognition result including a plurality of candidates for the object to be recognized by carrying out a template matching process on the image; a candidate exclusion processing unit that excludes a candidate that meets a predetermined condition by generating, for each of the plurality of candidates, a binary image of the object to be recognized, and finding a degree of overlap of each candidate using the binary image; and a recognition result output unit that outputs a candidate that remains without being excluded as a recognition result.