The Research Of Mean Shift Segmentation Algorithm For Medical Ultrasound Image Based On Texture

This article focuses on the Mean Shift algorithm ultrasound image segmentation method based on the texture information. The main research contents of the article are as follows:(1) Texture is one important characteristic of images. We can achieve the purposes of identification and classifications of images through distinguishing different textures. Ultrasound images have rich texture information which makes our work feasible. Many articles propose many analytical methods based on texture analysis, Gabor algorithm is a more widely used methods based on texture analysis in recent years. This paper introduced the application and development of texture analysis based on Gabor algorithm, and improved this method to extract the texture characteristics of ultrasonic images. Experimental results show that accurate segmentation the image segmentation based on the above method is more accurate and distinguishable.(2) Mean Shift, which is an adaptive peak search method without parametric, is of simple calculation and fast convergence. However, Mean Shift can easily produce wrong segmentation boundaries and lose partial details when it is applied to the ultrasonic image segmentation. We proposed an automatic threshold calculation method combined with matching bandwidth based on texture information above. By image smoothing and simple threshold clustering, Mean Shift algorithm resulted in a more accurate segmentation which confirmed the advantages of this method.(3) Finally, this paper shows the construction and integration of the proposed segmentation algorithm simulation interface. Development and integration of medical ultrasound image segmentation platform is developed based on Matlab7.0 with powerful image processing toolbox and simulation function, the platform interface can work-friendly. Simulation interface displayed the results of image texture extraction and segmentations, and users can visually acquire the improved results of algorithm and the comparison with the traditional algorithms.