Automatic recognition and classifier of online parts based on machine vision

摘要

In order to call correct NC program automatically, real-time for corresponding online parts in the flexible manufacturing system (FMS), a new automatic recognition and classifier system based on machine vision was developed. In the image pre-processing, to make the extraction of image edge-detection better, a new re-filter, consisting of three steps-Gauss linear smoothness filter, sharpening, Median Filter, was first introduced. Then, Canny edge detection algorithm was adopted. Moreover, comparing with the most existing classification methods, such as Nearest Neighbor, Bayesian, Off-Line computations and so on, a new classification algorithm, Two Steps Shape Classification, was proposed. Using a Radial Feature Token (RFT), which functions as the ALISA Shape Module in the Adaptive Learning Image and Signal Analysis (ALISA) system hierarchy. Experimental results confirm that the image processing algorithm is effective and useful for real-timely recognizing and classifying online parts in the FMS.