Self-adaptive Image Segmentation Method Based on Hilbert Scan and Wavelet Transform
基于Hilbert扫描和小波变换的自适应图像分割

Threshold selection is the critical process of image threshold segmentation method.Considering the neglect of spatial correlation of image pixels in current threshold segmentation methods,we propose to combine Hilbert image scan with wavelet transform to obtain a continuous and smooth threshold curve,and then propose a local self-adaptive threshold method in this paper.Firstly,the image is translated into 1D Hilbert order via Hilbert image scan method;Secondly,the curve of the developing trend of the Hilbert order is obtained by the multi-resolution analysis using wavelet transform.Furthermore this curve is chosen as self-adaptive threshold and the Hilbert order binarization is realized.Lastly,the binaried Hilbert order is translated into 2D image using the reverse Hilbert matrix scan and the image segmentation is achieved.The threshold curve achieved by the above-mentioned method is able to self-adaptively adjust along with neighborhood property,and reflects the developing trend of grayscale information in current image region.So the present algorithm preserves the local information and the relativity of adjacency pixels in the image.Moreover,it also improves the efficiency of image segmentation.Experiments indicate that the proposed method is an extraordinary effective image segmentation technique.It is with very good performance and immunes to noise.