Sub-pixel Edge Detection And Automatic Detection Of Geometrical Features

The workpiece contour is generally composed of the linear, circular and ellipticalarcs..Therefore correctly detected lines and arcs of contour is the key issue,in the imageprocessing.In this paper, digital image processing technology to identify workpiececontour geometry, which has a non-contact technique, flexibility, high precision andoutstanding handling characteristics.First of all，the definition of the edge and the edge detection algorithm arediscussed.In order to overcome the traditional low-pixel edge positioning accuracy,sensitivity to noise and other shortcomings.The sub-pixel edge detection algorithmbased on radial basis function interpolation,sub-pixel edge detection algorithm based onGaussian curve fitting and sub-pixel edge detection algorithm based on density functionare discussed.Compared test results to analysis the advantages and disadvantages ofthese sub-pixel edge detection algorithms.Then, contour recognition and segmentation algorithm is discussed.In order toidentify the geometric characteristics of the workpiece contour, this paper discussesmanually identify contour algorithm, semi-automatic contour recognition algorithm,automatic identification contour algorithm and their steps.It goes on to discuss the linear,circular and elliptical arc edge fitting algorithm.Finally and most importantly,for applied research, a software system to identify thecontour is developed.The test proved that the software system is able to identify theimage contour of the workpiece, in the case with noise.