Research On High Resolution Woodland Aerial Image Sequences Mosaic

Woodland survey mainly relies on ground investigation and remote sensing image.In recent years,unmanned of the Unmanned Aerial Vehicle(UAV)led to a rapid development,and it has been widely used in geographic information.Mosaicking the image sequence into a large panorama is one of the purposes of UAV image processing.In this thesis,we study the key steps in the stitching method of woodland aerial image to obtain high quality regional panorama and at the same time ensure the accuracy and stability.UAV image sequence stitching is based on image stitching technology.In this thesis,on the basis of analyzing woodland image in detail,we firstly calculate lacunarity value from image blocks to find several proper blocks which to be extracted the SURF registration points instead of the whole image.Additionally,some improved methods for feature matching was proposed.The constraint of orientation of feature makes contributions to accuracy of feature matching.In addition,we experimented with poisson fusion,the optimal seam line and the thin-plate spline(TPS)translation on image fusion and applied in our method to get a better result.Finally,compared with traditional SURF and down-sample stitching method,the proposed method gave a high-resolution panorama with excellent visual effects and high accuracy.