A Fast Image Recognition Method Based on the Combination of Extension and Subpixel Matrix

Abstract

The invention provides a fast image recognition method based on the combination of expansion and sub-pixel matrix. First, all images are preprocessed to construct a candidate image database; then an expansion matrix and a sub-pixel matrix of the candidate image database are constructed; and then the original pixel matrix is ​​constructed. A classifier, according to which all images in the image database are stored in a tree structure; then the image to be recognized is preprocessed to form the original pixel matrix of the image to be recognized; and the expansion matrix, sub-pixel matrix corresponding to the original pixel matrix of the image to be recognized is constructed pixel matrix; finally, the to-be-recognized image and the image in the candidate image database are gradually matched according to the extended matrix, the sub-pixel matrix, and the original pixel matrix, so as to quickly identify the result. The invention performs feature matching based on the image expansion matrix and sub-pixel matrix, gradually narrows the scope of the target database, greatly improves the information search matching degree and search efficiency, saves the matching time, and improves the accuracy of image recognition.

Description

technical field [0001] The invention relates to a fast image recognition method, in particular to a fast image recognition method based on the combination of expansion and sub-pixel matrix. Background technique [0002] The recognition of Chinese characters, numbers, alphabetic words, fingerprints, faces, etc. is a common and critical technology in life. At present, Chinese characters, numbers, alphabetic words, fingerprints, faces, etc. are used in many fields such as document retrieval, new construction and parcel sorting in the logistics industry, identification of various documents, rapid information entry, security inspection, and fingerprint attendance. Image recognition technology. [0003] The recognition of Chinese characters, numbers, alphabet words, fingerprints, faces, etc. generally includes several parts such as information collection, information analysis and processing, and information classification and discrimination. Commonly used methods are template ma...

AI Technical Summary

Problems solved by technology

Commonly used methods include template matching method and geometric feature extraction method. These methods have the defects of low recognition efficiency and insufficient recognition accuracy.

Images