基于改进尺度不变特征变换的手术室多视点图像拼接算法

目的为了提高手术室环境下拼接后全景图像的主观质量,本文提出一种基于改进尺度不变特征变换(scale invariant feature transform,SIFT)的手术室多视点图像拼接算法。方法基于SIFT特征点匹配算法,依据手术室采集设备为平行摄像机且获取的两视点高关注区域重叠较宽泛的特点,提出采用局部加权融合的拼接算法消除拼接重影,并在局部加权融合算法下,对参与图像空间变换的特征点在垂直方向增加约束,筛选出正确匹配特征点对以消除拼接折痕。结果此算法解决了手术室拼接图像中手术衣的袖子、手术剪和手套的重影及治疗巾位置拼缝不对齐的问题,即消除了术野高关注度下的拼接重影及误匹配带来的拼接折痕,保证了两幅图像拼缝后的平滑过渡。结论基于改进SIFT特征点匹配的算法获得了较好的主观手术画面的拼接质量。

Multi-view image mosaic algorithm based on improved scale invariant feature transform in operation room application

Objective In order to improve the subjective quality of the stitched panoramic image taken in the operation room,a multi-view image mosaic algorithm based on improved scale invariant feature transform ( SIFT) is proposed. Methods An algorithm with local weighted fusion was proposed to eliminate splicing ghosting phenomenon . Moreover,to eliminate stitching crease phenomenon in stitched images,new constraints were proposed and applied in vertical image space transformation to select the accurate matching feature points. Results The proposed algorithm solved the problem of mosaic in the stitched images,for example,the quilting appeared at the sleeve of the surgical gown, the operation shears, the gloves and uneven treatment towel position. Conclusions Better subjective quality of the stitched panoramic operation images is achieved by using the improved algorithm.