Research IntoAutomatic Generation Technology Of DSM For LineArray CCD Satellite Images

With the development of space technology and sensor technology, the monopolyof frame perspective camera has been broken, replaced by linear CCD sensors whichbecome one of the most effective ways to observe earth. The process of linear arrayCCD images have become a hot and difficult problems of the current study. In thispaper, automatic generation technology of DSM for linear array CCD images is themain content, around the image matching methods, epipolar theory and threedimensional reconstruction algorithm to have a more systematic theoretical research.And programming based on Visual C++2010to achieve automatic extraction of DSM.The main contents are as follows:1. Compared to frame perspective camera, the imaging principle and thephotogrammetry schematic of line array CCD sensor are introduced. And based on theRational Function Model, the paper describe the three-dimensional reconstructionalgorithm.2. Comparative analysis of KAZE and SIFT algorithms, the result is that thefeature points extracted by KAZE algorithm are more stable, which can be easy toremove the mismatching points. The KAZE algorithm can be prioritized as thematching algorithm for remote sensing images of detail and texture lost.3. Epipolar images can be generated by the projection track method. The methodof self-checking about epipolar images by image matching is presented in the paper tojudge the epipolar images’availability and whether to compensate RPC.4.Analysis of the method and key technologies of automatically generating DSMby dense matching for the stereo line CCD array images. The method of pyramidmatching one by one of epipolar images and the two-way matching of dynamicwindow transforming matching based on coefficient is presented to achieve gridpoints’ matching.Using the ZY-3images as the experiment data to indicate that those methodshave good performance.