基于区域蛙跳搜索与轮廓匹配的显微图像拼接

Microscopic Image Stitching Based on Regional Frog Leaping Search Algorithm and Image Contour Matching

为解决传统显微图像拼接中产生的几何畸变和错位,及特征稀少造成的正确匹配率低、时效性差等问题,提出基于区域蛙跳搜索和图像轮廓匹配的拼接算法。提取连续采集且有重叠区域的图像轮廓曲线;引入轮廓线索感知相似度和均方误差距离,计算图像轮廓曲线间的相似度或曲线离散距离,并将其作为匹配的衡量指标;在决策域内采用区域蛙跳算法更新鸣叫分贝和蛙跳策略,搜索图像轮廓最优匹配,实现图像快速精确的拼接。结果表明,所提算法不仅具有较高的拼接精度和较强的稳健性,还减小了其简化匹配策略的计算量,具有较强的时效性。

To solve geometric distortion and misalignment generated in the traditional microscopic image stitching process and avoid low matching rate and high computational burden caused by rare features,a stitching algorithm based on the regional frog leaping search algorithm and image contour matching is proposed.First,the image contour curves of continuous acquisition and overlapping regions were extracted.Then,the clue-aware trajectory similarity or mean square error distance was introduced to calculate the similarity and discrete distance between image contour curves.Finally,the regional frog leaping algorithm was used in the decision domain,and by updating the tweet decibel and frog leaping strategy,optimal contour matching for the image was searched.Consequently,the images were stitched quickly and accurately.Experimental results verify that the proposed algorithm demonstrates high stitching precision and strong robustness.In addition,the simplified matching strategy reduces the amount of calculation and exhibits strong timeliness.