The Research Of Remote Sensing Image Classification Based On K-Means And Image Transmissionn System

In order to send message unaffected by factors like network, signal,security and so on, our laboratory cooperated with the military, designed and developed a WNTS(wireless network transmission system). The WNTS can send encrypted message by RF signal through FPGA. The WNTS can not only comminucate with each other in normal conditions,but also can be used in field rescure, strategy, direction and so on. The WNTS was consist of wireless platform and application platform. Based on the system, the process of remote sensing image denoising and remote sensing image classification were studied. Finally, we sended the classified remote sensing images by the wireless network transmission system and we implemented a system for the transmission of classified remote sensing images. There are mainly four aspects in this article:1) Design and implement application platform in wireless network transmission system, design the swip communicating protocol between wireless platform and application platform, implement the transmission of message between wireless platforms and design the process of transmission.2) Analysis the noise type in remote sensing images and use median filter and wavelet threshold denoising to remove noise in remote sensing image. Propose an improved double-threshold wavelet threshold denoising function to overcome the shortcomings of traditional wavelet threshold denoising function.3) Analysis and compare the common methods for image classification, use k-means algorithm to classify remote sensing images. Consider the defection of traditional k-means algorithm,propose an improved k-means algorithm which can adptively determines the number of clusters and optimizes the initial cluster centers.4) Integrate remote sensing image classification function into the wireless network transmission system, implement a classified remote sensing images transmission system.In this paper, we use PSNR(Peak Signal to Noise Ratio) as the objective evaluation standard in image denoising, we compare the image denoising effect between traditional wavelet threshold denoising method and the improved wavelet threshold denoising method by matlab simulation. Matlab simulation results show that improved wavelet threshold denoising method is better than traditional wavelet threshold denoising method in remote sensing image denoising. The improved k-means algorithm is better than traditional k-means algorithm in remote sensing image classification as well.