【摘要】 图像拼接过程中存在着视角、颜色、纹理信息等差异，这些差异严重影响了图像拼接质量。针对此问题，论文提出了一种改进最优拼接线的图像拼接算法。在图像配准方面，基于超像素分割算法对局部单应性模型进行改进，提高了图像配准的精度，削弱了视差对图像拼接过程的影响；在接缝切割方面，基于原有平滑项中考虑了色差与梯度差信息，在平滑项中添加了显著性感知权重与纹理性信息。实验证明，该算法提高了图像配准的精度，保护了视觉显著物体结构，避免了错位、重影现象的发生，能够拼接生成高清自然的宽视域图像。更多还原

【Abstract】 In the process of image mosaic,there are differences in perspective,color and texture information,which seriously affect the quality of image mosaic. To solve this problem,this paper proposes an image stitching algorithm based on improved optimal stitching. In the aspect of image registration,the local homography model is improved based on the super-pixel segmentation algorithm,which improves the accuracy of image registration and weakens the influence of parallax on the image mosaic process. In the aspect of seam cutting,based on the consideration of color difference and gradient difference information in the original smoothing item,significant perceptual weight and texture information are added to the smoothing item. Experiments show that the algorithm improves the accuracy of image registration,protects the structure of visually significant objects,avoids the occurrence of dislocation and ghosting,and can splice high-definition natural wide field of view images.更多还原