Research On The Technology Of Mosaic Of Aviation Remote Sensing Image From UAV

The Unmanned Aerial Vehicle (UAV) Aerial Remote Sensing (ARS) system has the advantages of high image resolution, real-time image transmission, operation in high-risk regions, low cost, mobility and flexibility, etc. UAVARS system is suitable for the real-time acquisition of high-resolution RS data at low altitude and plays a role which is irreplaceable by large-scale RS system in territorial, constructional, catastrophic and military RS monitoring. When carrying out missions of RS monitoring, UAV needs to transmit by real-time the obtained images and status data, therefore the functions of automatic and rapid acquisition, compression, transmission, processing, display and storage are required of the ARS system. Among those functions, accuracy, real-time performance and visibility of the RS image processing are important preconditions of the effective utilization of the UAV. Due to the limited performance of the existing imaging devices, the existing ARS imaging system cannot obtain observed images of large area or high resolution, and the system needs to dynamically splice the obtained sequence RS images on line to improve the information-obtained capability of the RS images. Based on the specific requirements for the application of the aerial UAVRS, this paper did the following researches on ARS image-splicing technology:(1) According to the interior and exterior position elements of the imaging of the UAV RS images, right angled spatial transformation and method of quadratic linear interpolation are used to correct the RS images. According to the obtained flight-status parameters of the UAV, it is calculated the range of the overlapped-area image of two ARS quick-images taken continuously by the UAV.(2) A parallel geometrical correction algorithm was provided based on distributed memory systems. In the algorithm, each processor calculates the corresponding area in the target image for the local sub input image, and does resampling for this area. This makes all of data needed be in local memory and no communication happens during parallel computing. Closed line segments connected end to end with each other are used to represent the ideal edge of each sub output image approximately when calculating local output area and a data structure is put forward to save irregular sub output images.(3) Based on the visual characteristics curve of human-eye brightness, Combine the characteristics of Wavelet and Curvelet, the self-adaptive enhanced processing of the UAV ARS quick-images are realized.(4) A new edge detector is proposed. Combined Wavelet and Canny detector, it can preserve the large-scale edges and ignore the sharp textures. It is suit for image registration.(5) Image matching includes two procedures, i.e. rough matching and fine matching. In rough matching, the area coverage of overlapped regions between two consecutive images to be stitched is determined approximately at first. Both of the overlapped regions of the two images are processed by wavelet transform and Canny Operator, and then the edge of the images has been extracted. Obtained the match point used regional matching method. In the fine matching, the sequential similarity detection algorithm (SSDA) is adopted to perform matching computation in the some small regions near the positions got in the rough matching, and then the relative position offsets in X-orientation and Y-orientation between the two consecutive images are got. Based on the result of the image matching, the two images are stitched.(6) Based on the analysis of human-eye color vision characteristics, an anti-brightness-disturbance color difference measuring method for color images is put forward. By using analysis of color similarity and introducing covariance matrix calculation, a method for extracting the characteristics template of color images is given. By using least square method, it is established the function curve of brightness transformation between two color images, and the gamma correction is realized based on the reference image brightness-distribution. On the RS image processing algorithm put forward by this paper, the simulating programming is realized, the feasibility of the algorithm is verified, and the design of the dynamic splicing software for UAVARS images is accomplished.