A Study On Seam-line Removal Under Mosaicking Of Remote Sensing Images

Seam-line removal is a problem which needs to be solved in the processing of image mosaic, because theseam-line directly affects visual quality, as well as the further application of the image. We improved theexisting methods on the foundation of researching on the algorithms of seam-line removal and determinationof the seam-line. The paper's main works are as follows:1. In the dissertation, we summarized the existing seam-line removal algorithms in detail, and describedthe main principles and applicability of each algorithm, which are the foundation of the further research.2. We proposed a seam-line determination method depending on the principles of optimal seam-linedetermination algorithm and seam-line determination based on gray correlative. The seam-line determinationalgorithm based on Cross Correlation Function defined the Cross Correlation Function as the exclusive texturedifference degree, as well as expanded searching width from three pixels to five pixels. The experimentshowed that determination of the seam-line based on CCF method has advantages over the traditionalseam-line determination.3. We analyzed the principles of solving differential equation in depth, because of the shortcomings ofwhich the seam-line removal algorithm based on gradient field can not correct the entire image, the mixedboundary conditions for the seam-line removal based on gradient field algorithms is proposed. We elicited thatthe mixed boundary conditions for the seam-line removal algorithm based on gradient field is better than theoriginal algorithm by analyzing and comparing experiment data.4. Aimed at the disadvantage of the application of seam-line removal algorithm based on gradient fieldconsume much time, image segmentation based on linear quad-tree was introduced into seam-line removalalgorithm. After linear quad-tree image segmentation, the leaf node represents the entire area, the elementsamount of the linear equations iterative reduced. The proposed algorithm eliminated the seam-line as well asimproved the efficiency of the original algorithm and reduced memory consumption.