一种加权局部熵羽化的图像拼接模型

Model of Weighted Local Entropy Feather Image Mosaic

图像拼接技术能多幅背景相似的图像合成为一张全景图像,也可以补充某些图像缺失的部分,但是目前常用的拼接算法还存在准确性不高、鲁棒性不强的缺陷,为此本文提出了一种加权局部熵羽化的图像拼接模型。首先对图像的局部熵进行羽化,然后采用一元线性回归的方法对邻域像素的灰度图像进行相似度的优化,最后对图像的局部熵进行加权,以突出图像中领域像素之间的块相似度,降低羽化的痕迹。模型的对比仿真结果表明,本文提出的算法相比较SIFT算法,对于动态图像的拼接具有更高的准确性,并且抗噪性能更加,具有较好的鲁棒性。

Image stitching technology can synthesize multiple images with similar background into a panoramic image,and can also supplement the missing part of some images. However,the commonly used stitching algorithms still have the drawbacks of low accuracy and poor robustness. To this end,This paper presents a weighted local entropy feather image mosaic model. Firstly,the local entropy of the image is feathered,and then the unary linear regression method is used to optimize the similarity of the gray image of the neighborhood pixels. Finally,the local entropy of the image is weighted to highlight the similarity between the blocks Degree,reduce the traces of emergence. The simulation results of the model show that compared with the SIFT algorithm,the proposed algorithm has higher accuracy for splicing the moving images and has better anti-noise performance and better robustness.