An automatic shoe-groove feature extraction method based on robot and structural laser scanning

Abstract

In order to improve the operational efficiency of robot-based shoe manufacturing, a method of shoe-groove tracking based on industrial robot is presented in the article. First, side surface of a shoe upper with a sole is scanned with a laser scanning device. The presented approach mainly consists of two steps: reconstruction of three-dimensional point cloud and feature curve extraction. It is difficult to extract the closed groove curve on shoe surface. We propose an innovative method to simplify the feature extraction through projecting geometric information from three dimension to two dimension, which is convenient to identify longest groove feature line in two-dimensional space. After detecting the two-dimensional groove line, we back project it to three-dimensional space to identify the three-dimensional thick groove point set. Finally, we thin and fit the groove curve into a trackable sequential curve. The experimental results show that the proposed system can effectively detect the shoe groove and generate trackable sequential curve. We also simulate the robot tracking process in a virtual environment to demonstrate the effectiveness of the presented method.