基于小波变换的遥感图像快速拼接方法

Fast Image Mosaic Based on Wavelet Transform for Remote Sensing

基于图像小波变换与低频区域特征匹配的拼接方法,实现无人机序列遥感图像的快速动态拼接.根据无人机遥感图像成像的内、外方位元素,采用直角空间变换及二次线性插补方法,实现了遥感图像校正.小波变换提取低频图像,在此图像区域中搜索和提取特征模板,然后利用序贯相似性检测法进行匹配计算.根据匹配结果,实现两幅图像的拼接.仿真实验结果表明,所提出的拼接方法具有较好的实时性和拼接精度.

A method of serial fast image mosaic was provided from UAV （unmanned air vehicle）, based put forward for the remotely sensed images on wavelet transform and matching of low- frequency images. According to the interior and exterior azimuthal elements of the remotely sensed imaging by UAV, the rectangular spatial transformation and second order linear interpolation are used to correct the images. The method proposed uses the wavelet transform to extract the low-frequency sub-images and search the relevant area, thus obtaining the characteristic image as template. Then, the SSDA （sequential similarity detection algorithm） is introduced to perform the image matching computation and after analyzing the result, two images are mosaiked together. Simulation results showed that the algorithm presented greatly improves the operation speed with high precision and that it can be applied to real-time mosaiking of serial images obtained from UAV.