Research On Image Measurement Technology For Cutting Tool Geometric Parameter

Rapid and accurate offline measurement of the cutting tool parameters could effectivelyimprove the machining quality and efficiency of NC machine. Compared with the mechanicalor projection methods, the advantages of the method based on image measuring technologywere non-contact, high precision, automation and efficency.For this reasion an image measurement system for the cutting tool parameters wasdesigned which consisted of three parts: the optical imaging system, the mechanical platformand the image measurement software. The working principle of this system was as follows.The tool was projected onto the CCD sensor inside the camera through the lens in thebacklight mode and photoelectrically translated into the tool image.Then the image wasprocessed by the image measurement software and the result was fused with the data from themechanical platform.Finally the tool parameters were obtained.In this paper，several key algorithms of the image measurement software which was thecore of the system were studied just as below and the design idea of the measure program wasalso proposed based on the algorithms.It was much faster for the sub-pixel precise thresholding based on the combination of thenormal threshold operation and the bilinear interpolation than the classic sub-pixel edgedetection algorithms in extracting the contour of a high contrast tool image accurately.The locations of the corresponding dividing points produced by the Ramer algorithmwere amended, which led to a more accurate approximation of the contour.The segmentationof contours according to primitive property was achieved by merging the qualifyingadjacent contours into arc.The Tukey weight function was introduced to the line and arc fitting algorithms so thatthe improved fitting algorithms performed a strong suppression effect for the lagre outlierpoints. Thus the measurement accuracy was indirectly improved.Finally, the stability and repetitive accuracy of the system were tested through theexperiment.The error sources were analyzed and some improvements were proposed.