Geometric Descriptor-Based Local Contour Point Set Matching for Cross-Section Profile Measurement of Automotive Sealing Strips

In vision-based measurement, there is a lack of effective methods for local contour matching of flexible automotive rubber product contour due to its flexibility. In this paper, we propose a novel local contour point set matching method based on a geometric descriptor. A geometric descriptor of a local point set includes shape properties of the point set and position information. In local matching process, the geometric descriptor of a standard point set is extracted and a tolerable geometric descriptor for a desired point set is also generated through adding more factors to the standard descriptor. According to the generated tolerable descriptor, an angle similarity optimization searching is used to find the best matched local point set within a constrained position limit. This method has been implemented in a sealing strip cross-section measurement system, and several cross-section profiles of automotive sealing strips have been tested. The results show the stability and effectiveness of the proposed method and this method can lead to an automated measurement for flexible products.