Rapid matching of line traces based on laser detection signal

It's usually not easy to compare line traces quickly and quantitatively with the current image processing and three-dimensional scanning methods. In order to deal with these kinds of problems, a similarity matching algorithm for laser detection signals of line traces is proposed. The proposed algorithm does section leveling for shear plant first, and then applies histogram to define the abnormal fluctuation data between the neighboring points, utilizes a K-Means clustering to eliminate the abnormal data. After that, locally weighted scatterplot smoothing algorithm is used to performs noise reduction on the surface signals of line traces that are picked up by a laser displacement sensor; followed by the identification of the features of ‘thick and consistent’ trends in the signal data. Then the feature vectors are quantized by cosine vector curve fitting to individually calculate the spatial distance of the sample. Results from similarity comparison tests of actual cutting traces validate the accuracy and validity of proposed method.