Biomedical image feature extraction method

Abstract

The invention discloses a biomedical image feature extraction method. The method comprises the first step of extracting lower-layer visual features (grey features, textural features and edge features) of a biomedical image, the second step of extracting a text region from the biomedical image to obtain pixel information and position information of the text region, the third step of extracting global features from the text region of the biomedical image, the fourth step of dividing the image into sub-regions through a sub-region division method, analyzing text region distribution of all the sub-regions and extracting spatial distribution features of the image, and the fifth step of conducting feature fusion on the three kinds of image features to obtain final image visual features. By means of the method, the essential properties of the biomedical image can be described comprehensively by analyzing lower-layer visual information and text region information of the image, so the classification accuracy of the biomedical image is effectively improved, and the effect is better when the method is used for identifying flow charts and graphs.

Description

technical field [0001] The invention relates to the field of digital image processing, in particular to a biomedical image feature extraction method. Background technique [0002] With the development of computer multimedia technology and machine learning technology, the application of image database is becoming more and more extensive. How to effectively establish, manage and make full use of image information database resources to realize intelligent and efficient recognition and retrieval of images has always been a hot issue concerned by scientific researchers at home and abroad. In scientific research, national defense, industrial production, aerospace, It has a wide range of applications in biomedicine, traffic monitoring and other fields. Faced with the characteristics of large amount of biomedical image data, multi-modality, complex imaging, and strong pertinence, feature extraction technology is constantly facing new opportunities and challenges [0003] Image fea...

AI Technical Summary

Problems solved by technology

[0005] After analyzing the existing technical literature, scholars such as R.Barry and L.Minsuk wrote in "R.Barry, L.Minsuk, Exploring text and image features to classify images in bioscience literature, Proceedings of the Workshop on Linking Natural Language Processing and Biology, 2003, 73-80", proposed a feature extraction method based on the combination of text and vision. This method mainly extracts visual features (features based on histograms, 8-connected region features) and image captions based on biomedical images. Text features, and realize the classification of biomedical images, but these features have low classification and recognition accuracy for images, and ignore the information of text in images

Images