Method and device for identifying workpiece based on binary descriptor

Abstract

The invention discloses a method for identifying a workpiece based on a binary descriptor, comprising the following steps: 1) extracting characteristics points by using the Fast Hessian characteristics detection operator; 2) respectively mapping the characteristics point information extracted by the Fast Hessian characteristics detection operator and the grayscale information of the pixels in each sub-region to a circular sampling mode to construct a workpiece characteristics descriptor; 3) using a cascade-type matching algorithm to the obtained workpiece characteristics descriptor and the template characteristics descriptor in a template base and using the nearest neighbor ratio for Hamming distance matching; obtaining the initially matched pairs and making statistics on the number of the initially matched pairs; 4) using a random sampling consistency algorithm, excluding the erroneously matched pairs from the initially matched pairs and obtaining the number of the correctly matched pairs; and 5) according to the matched pairs, calculating the matched score so as to obtain the workpiece recognition result. In the invention, a binary characteristics description algorithm combining the FREAK descriptor and the Fast Hessian characteristics detection algorithm is used to realize the fast and accurate recognition of a workpiece.

Description

technical field [0001] The invention relates to a workpiece recognition device and method, in particular to a workpiece recognition method and device based on a binary descriptor. Background technique [0002] Workpiece recognition is a typical application of computer vision technology in the field of industrial production, and is an important part of production automation and intelligence. At present, the research in the field of workpiece recognition includes methods based on image contours, methods based on image invariant moments, methods based on local features of images and workpiece recognition methods based on binary features. [0003] Yuan Anfu et al. proposed a part recognition algorithm based on SURF (Speed ​​Up Robust Feature) features. In this scheme, an industrial CCD (charge-coupled device) camera is used to obtain the workpiece image first, and then preprocessing operations are performed on the workpiece image, mainly including image enhancement, median filt...

AI Technical Summary

Problems solved by technology

The above method uses the FREAK descriptor, which has a high degree of overlap in the smooth range of sampling points, excessive redundant information, and poor scale invariance; the FREAK descriptor only uses the comparison results of a single neighborhood of sampling points to form a binary descriptor. Lack of hierarchical information; in addition, FREAK relies on the main direction technology to achieve its rotation invariance, which is less robust

[0005] Although the existing methods can complete the basic workpiece recognition, there is a problem of relying too much on the main direction of the feature points to achieve rotation invariance, resulting in poor robustness of the algorithm to rotation

Images