【摘要】 针对传统的基于特征的图像拼接算法不能很好地解决大运动物体造成的场景误匹配问题,提出了一种基于特征的包含大运动物体的图像拼接算法。在SIFT提取特征点并匹配的基础上,通过随机采样一致性算法去除误匹配点,并通过观察特征连接线角度的波动值,自适应判断场景中是否出现大运动物体以及场景匹配是否出现错误,最后通过改进的加权方法实现图像融合,很好地解决了运动物体的干扰。试验表明,该方法有良好的效果和鲁棒性。更多还原

【Abstract】 For traditional feature-based image stitching algorithm can not solve the question of the mismatch of the scene caused by large moving objects,a feature-based algorithm for image mosaics of scenes with big moving object was proposed.By utilizing SIFT to extract feature points,then optimized random sample consensus was used to filter the matching points in order to eliminate mismatched ones,and observe the feature angle of lines to judge whether large moving objects were in the scene.At last,improved weighted function was used to stitch the images.The experiment showed that the method has good effects and robustness.更多还原