A Texture Classification Method Based on Extreme Learning Machine

Abstract

The invention discloses a texture classification method based on an extreme learning machine. The method includes the following steps that feature extraction is carried out on a known texture image sample to obtain a texture feature vector; the extreme learning machine serves as a base classifier, the texture feature vector serves as an input element of the extreme learning machine, the base classifier is trained through a training sample set, and a classification model is built; feature extraction is carried out on an unknown texture image, output vectors of a plurality of base classifiers are obtained according to a built dynamical model; the output vectors obtained from the unknown texture image are fused through the dynamical model, and the unknown texture image is recognized by means of the maximum decision rule. The method achieves automatic classification and recognition of the texture images, can obtain higher classification process, improves work efficiency and stability, has the advantages of being high in precision, high in speed, high in stability and the like, and can be used for automatic detection of texture images.

Description

technical field [0001] The invention relates to the intersecting fields of computer pattern recognition and texture images, in particular to a texture classification method based on an extreme learning machine. Background technique [0002] Texture recognition is an important research content in the field of computer vision and pattern recognition, and is widely used in agriculture, industry, medical treatment, military and other fields, such as the classification of different terrains and landforms in remote sensing systems, face recognition, and medical science in biomedical image analysis systems. Detection of diseased tissue in images, etc. At present, the research of texture recognition technology has made some progress, which provides a good platform and opportunity for the development of many fields, and attracts more and more researchers to further study and explore in related fields. [0003] Texture recognition is to determine the category attribute of an unknown ...

AI Technical Summary

Problems solved by technology

Aiming at the shortcomings of the above-mentioned algorithms, such as low classification accuracy rate, large amount of calculation and lack of theoretical support, some researchers proposed the extreme learning machine classification method (ELM), which is a new type of single hidden layer feedforward neural network developed in recent years. Network; different from the traditional method, the extreme learning machine randomly selects the hidden layer neurons in the network, and the output layer weight of the network can be obtained by analytical method, which has many advantages such as fast learning speed and good generalization ability.

However, since the input weight and hidden layer deviation between the input layer and the hidden layer of the traditional extreme learning machine are randomly assigned, it is easy to cause the output result to be unstable, which limits the method to a large extent in practical engineering. Applications

Images