Chinese character recognition of e-commerce platform pictures

Design is introduced by using of wavelet denoising, canny operator, genetic algorithm and other related methods to achieve the detection of characters in the picture. Then adopt a series of methods about normalization, feature extraction and so on to get the preprocessed character pixel feature, and then use the feature extraction method of stroke extraction to analyze the feature vector of the character. Finally, the resulting feature value is transferred to the existing OCR font library for comparison, so as to realize the character recognition in the picture.