| This paper analysed various image segmentation algorithms to find the suitable segmentation method of the foundation nephogram, and by using this method to make the automatic segmentation come true, which has certain theoretical significance and application value.First, it analysed and compared the Threshold segmentation method,Edge detection segmentation algorithm and Clustering image segmentation algorithm. The key point focused on the basic theories of wavelet transform and the application in image segmentation. Then, combining the characteristics of foundation nephogram it put forward a segmentation algorithm based on subdividing the wavelet and the Kmeans clustering. The Algorithm process was as follows:first, pretreating the original nephogram (is noise removing, reinforcing); secondly, exchanging the nephogram by subdividing the wavelet which could obtain a small image that contain the most information of the nephogram; next, using the Kmeans clustering to segment the small image.; now the basic nephogram has been segmented, however, since there do exist some over-segmentation, so taking the regional growth method to delete the over-segmentation; then, mapping the small image to the coordinate of the original nephogram; lastly, showing the boundary of segmentation image in the original grayscale image to clearly observe the segmentation effect.Following, it used the MATLAB to implement the algorithm. The result of the experiment showed that comparing the traditional image segmentation algorithm, this algorithm has stronger anti-noise ability, higher accuracy of edge detection, better edge consistency, comparatively more integrity image segmentation and the inspection of the weak edge is also effective. It could be used in the segment of the foundation nephogram.This paper designed a new segmentation system of the foundation nephogram by MATLAB, which provided nine segmentation methods and basing on the outcome to distinguish the cloud cover of foundation nephogram.. |
PDF Full Text Request