A novel fast image segmentation algorithm for large topographic maps

Image segmentation for topographic maps is challenging due to their low quality, high degrees of mixed and false coloring. Besides, many pixels cannot be explicitly separated from each other because of the fuzziness of their colors. Therefore the algorithms based on fuzzy theory are suitable to process such images utilizing their ability to deal with the blurring effect. However, there are still some problems of large-scale data, high time complexity and inaccurate classification. In order to overcome these problems, we propose a novel algorithm for segmenting large topographic maps based on the ideas of fuzzy theory, randomized sampling and multilevel image fusion. In this algorithm, the large topographic map is randomly sampled first. Then the optimal clustering centers are acquired by fuzzy C-means (FCM) clustering. Further, the map is segmented with the fuzzy classification method. Finally, multilevel image fusion is used to fuse the segmented images into the final segmentation maps. Randomized sampling is used to reduce the amount of data, and improve the efficiency of image segmentation. Multilevel fusion can make the final segmentation maps more accurate. The experiments show that our method has higher efficiency and accuracy than the traditional ones. It provides a reliable image segmentation method for large topographic maps.