A method to distinguish true and false edge points of etched characters

Abstract

In order to remove the false edge points extracted by the Canny operator in etched character recognition, a method based on the simplified neighborhood feature and AdaBoost algorithm was proposed. In conventional neighborhood feature, canny edge points are taken as the center and neighborhood pixels are extracted as the feature. However, the dimension of the neighborhood feature rises with the square of the neighborhood size, which increases the number of patterns and the robustness of the feature to illumination is weak for the gray value is variable in different images. Therefore, to overcome the limits of neighborhood feature, simplified neighborhood feature takes sample of pixels in horizontal and vertical directions to simply the feature, and quantizes the gray values by analyzing the gray histogram. Considering this is an unbalanced problem, the number of false edge points is far less than true edge points, AdaBoost algorithm is used to classify SN feature. Result shows that SN feature can correctly classify edge points across change in illumination.