Workpiece Position Measurement with Segmented Edge Extraction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing methods for measuring the position of a workpiece from a search image face accuracy issues when the number of edges extracted is reduced, often due to noise, missing edges, or shadows, leading to decreased accuracy in position and orientation measurements.
Innovation Solution
An image processing method that includes a model acquisition step, a rough search step, an edge extraction step, and a close search step, where edge extraction regions are set and edges are fitted to measure the workpiece position, with the option to incorporate model edges in regions where extraction fails to maintain accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If template matching is used to search for workpiece position, then the method is simple and widely applicable, but the measurement accuracy decreases when edges are noisy, missing, or uneven
Solution Approach 1:
The patent segments the edge extraction process into multiple independent regions corresponding to different model edges. Each region is processed separately to extract edges, and the results are combined for fitting. This segmentation allows the system to handle missing or noisy edges in individual regions without compromising the overall measurement accuracy, as other regions can compensate.
Solution Approach 2:
The patent performs preliminary edge extraction in multiple specific regions before the final fitting operation. By pre-extracting edges in defined regions and storing them, the system prepares complete edge data in advance, ensuring that even if some edges are missing or noisy during actual measurement, the fitting can still proceed with sufficient data from previously extracted regions.
2Measurement precision
If circle fitting or ellipse fitting is performed after template matching, then position measurement accuracy improves for circular or elliptical workpieces, but the process becomes more complex
Solution Approach 1:
The patent divides the fitting process into segmented steps: first performing template matching to obtain a rough position, then extracting edges in multiple specific regions around the rough position, and finally performing circle or ellipse fitting only if sufficient edges are obtained. This segmented approach reduces complexity by conditionally applying fitting only when beneficial.
Solution Approach 2:
The patent extracts edges in more regions than the minimum required for fitting (excessive action). By setting multiple edge extraction regions that may overlap or extend beyond what is strictly necessary, the system ensures sufficient edge data is obtained for accurate fitting while maintaining a systematic and manageable process structure.
3Measurement precision
If shadows or other obstacles hide features in the captured image, then edge extraction fails in those regions, but the workpiece position still needs to be measured accurately
Solution Approach 1:
The patent segments the edge extraction into multiple independent regions, each corresponding to a specific model edge. When shadows or obstacles cause edge extraction failure in one region, the system can still extract edges from other unaffected regions. The fitting process then uses the available edges from multiple regions, compensating for missing data in individual regions.
Solution Approach 2:
The patent changes the parameter of edge extraction by performing it in multiple specific regions rather than attempting to extract all edges from the entire image at once. This regional approach allows the system to adapt to local conditions (such as shadows in specific areas) while maintaining overall edge extraction effectiveness through aggregation of results from multiple regions.
Data Source
AI summary
A model containing a plurality of model edges is acquired, and a position that matches the model is searched for within a search image. Within the search image, a plurality of edge extraction regions including the respective plurality of model edges when the model is moved to the position searched for are set, and an arithmetic operation of extracting an edge in each of the plurality of edge extraction regions is performed. If there is an edge extraction region where no edge has been able to be extracted, of the model moved to the position searched for, a model edge located in the edge extraction region where no edge has been able to be extracted is incorporated into an edge set, and fitting is performed to the edge set.