The Research On Change Detection Based On Fuzzy C-Means Algorithm In Remote Sensing Image

In order to strengthen land and resources management and promote sustainable economic development and overall social progress. It's urgent to dynamically monitor land and resources using a scientific approach. Remote sensing technology has the rapid, accurate, and short periodicity characters, so it has an obvious advantage in land and resources management. Researching a remote sensing satellite monitoring platform for monitoring land and resources has become a hotspot. Change detection is a key part in the remote sensing satellite monitoring platform, and has been extensively studied. On the basis of summarizing and comparing the popular change detection algorithms, we focuse on the change detection algorithm based on fuzzy c-means (FCM) algorithm in remote sensing image. In our research, we deploy a fast FCM algorithm to classify remote sensing images, then, use a multi-band integration mask method to do the change detection.The content and innovation in this paper are as follows:(1) A combination mountain function\density function initialization method is proposed, because in traditional FCM algorithm, the clustering performance is affected by initial centers, and the calculation burden is heavy in the process of clustering initialization. The combination mountain function\density function method not only can solve the inefficient problem in the initialization method based on density function, but also can effectively alleviate the problem, which the computation grows exponentially with the dimension of the object space, in the initialization method based on mountain function. The initial cluster centers of our initialization method are quasi-optimal, which can make the FCM algorithm fast convergence and can improve the accuracy of the FCM algorithm.(2) A fast FCM algorithm for remote sensing image clustering is proposed to solve the problem of heavy calculating burden in traditional FCM. In the fast FCM algorithm, we use the combination mountain function\density function method to determine initial centers. In order to solve the coincidence problem between cluster centers and data points in the efficient FCM with reduced time complexity, an improved efficient FCM algorithm was proposed, and we utilize this method to classify remote sensing image. The fast FCM algorithm can improve accuracy and efficiency in remote sensing image classification, and can further improve the performance of change detection.(3) A multi-band integration mask method for change detection was proposed, because the combination image enhancement/post-classification method always ignores useful information. In the multi-band integration mask method, firstly, comprehensive analysis every band in remote sensing image and form a multi-band integrated change mask. Secondly, overlaid the change mask onto the date 2 image and only those pixels that were detected as having changed are classified in date 2 image. Thirdly, compare the classification results of date 1 image and date 2 image. By this means, the change type and position information can be yielded. This method can reduce change detection errors and improve change detection accuracy. Experiments show that the multi-band integration change mask method is effective.