Study on the method of automatic measurement of flexible material processing path based on computer vision and wavelet

Straightness, roundness, and primitive angle error of contour are important indicators of evaluating path precision during flexible material path processing. As processing path is composed of small arc or small line segment primitives, also the deformation of the flexible material during the processing path, making the captured image of processing path not clear, the edge of processing image over local uneven gray, the pixels of boundaries between the processing path image edge and background organizations not obvious. In order to extract the flexible material path contour effectively, mosaic method for flexible material processing path image is studied, next fast positioning strategy is introduced, and then we puts forward the search algorithm which taking processing path corner search as the cut-in-point, designing slope angle curve of starting and terminal point of each primitive and conducting slope angle curve for multiple scales wavelet transform by regarding DB(4) as wavelet operator based on wavelet edge modulus maxima extract principle. By judging whether one point of the curve appears at wavelet transform extremum, it can be determined whether the point is a corner one. In order to accelerate wavelet transform computing speed, FPGA IP core is 8-tap transpose is used to design the decomposition and reconfigurable of DB(4). The total time consumed by IP core wavelet decomposition increased only 2.802% compared to the PC computation time; path angle relative error is 8%, and the average measurement time is 198.22 ms.