The Research Of Stereo Vision Technology For The Korean Pine Forest Based On The Field Server

The Korean pine is a kind of precious tree species and important support for the economy of the forest. In order to protect the Korean pine forest and increase the seed yield, we blend modern information means in the research for the forecast method of the Korean pine cone yield. Those means include field server and technology of stereo vision.This thesis introduced the research situation at home and abroad of the field server and the stereo vision technology, and especially studied the architecture of the field server, camera calibration and the stereo matching. The existing calibration methods and the stereo matching algorithms were analyzed and compared, and then improved the algorithms according to the environment characteristics of the Korean pine. It established a certain theoretical and practical foundation for better protection of the Korean pine and improving the economic benefits.This thesis designed the field server applied to the Korean pine forest according to the actual situation of the Korean pine forest. It was an aggregation formed by a variety of sensors, wireless network and processors, which was applied to store and process the data information. The field server realized the purpose that we can collect the information data with less damage and reducing the human factors.This thesis improved the Tsai two-stage method in order to improve the calibration accuracy because of the radial distortion considered only in the original traditional Tsai two-stage method. There were both radial distortion and tangential distortion in the new method, so the calibration result would be more accurate and more stable.The region matching algorithm's characteristics were a large amount of computation, easy to be affected by noise and appear false matches. Besides it can get the whole image disparity map and apply to the natural environment. However, the feature matching had the fast calculation, high precision, the certain ability of resisting noise and little ambiguity characteristics. But it didn't apply to the natural environment. In that condition, a improved matching algorithm was put forward according to the environment characteristics of the Korean pine. This algorithm combined with the region matching and the feature matching, learning from each other. So the reliability if the matching was enhanced and the amount of calculation was reduced.