A wavelet transform-based image segmentation method

Abstract

In order to retain the details of image and automatically determine the segmentation threshold of image segmentation, an image segmentation method based on wavelet transform is proposed in this paper. The detailed information of horizontal, vertical and diagonal directions can be obtained by using image decomposition based on wavelet transform theory. The proposed method is tested on different images and it has better visual effect than other traditional methods, such as maximum variance segmentation, bi-modal segmentation and histogram valley threshold segmentation. Among the three image quality evaluation indicators, the proposed method also has obvious advantages, especially in medical image segmentation. This will provide a new method for medical image segmentation in the future, and also provide a direction for further research of medical image segmentation methods.