Color image feature extraction method based on fusion of color features and convolutional neural network

Abstract

The invention discloses an image feature extraction method that combines color features and a convolutional neural network, which mainly solves the problems of incomplete color feature extraction and insensitive color feature extraction of the convolutional neural network in the prior art. The implementation steps are: (1) input color image; (2) obtain subject image; (3) obtain color feature vector; (4) obtain normalized convolutional neural network feature vector; (5) obtain fusion feature vector. The invention can simultaneously play the advantages of color feature extraction and convolutional neural network feature extraction, and can be used in the technical field of image processing.

Description

technical field [0001] The invention belongs to the technical field of image processing, and further relates to an image feature extraction method in the technical field of image classification that combines color features and convolutional neural networks. The color image features extracted by the invention can be used for classification of color images and retrieval of color picture databases. Background technique [0002] At present, color image feature extraction has become a very important research topic in the field of image processing technology. The feature extraction of color images has a wide range of applications, such as target recognition and detection, image clustering, big data retrieval and other fields. Due to the complex content and variety of color images, the feature emphases of different images are different, which makes the feature extraction of color images very challenging. [0003] The color feature extraction based on image color and the convoluti...

AI Technical Summary

Problems solved by technology

Although this method effectively uses the color histogram as a color feature extraction tool, it still has the disadvantage that the method ignores the influence of background and other irrelevant elements on the color feature extraction, and the extracted color features have huge background noise

Although this method has improved the convolutional neural network feature extraction method, there are still shortcomings in that the method does not involve the feature extraction of image color features in essence, and the insensitivity to color features is the current convolutional neural network. The biggest drawback of the feature extraction method

This method is not good for feature extraction of color images with similar content but color differences, and the obtained color image features are not ideal for classification and retrieval.

Images