节点文献

基于Forstner算子的血细胞显微图像拼接

STITCHING OF MICROSCOPIC IMAGES OF BLOOLD CELLS BASED ON FORSTNER OPERATOR

  • 推荐 CAJ下载
  • PDF下载
  • 不支持迅雷等下载工具,请取消加速工具后下载。

【作者】 张从鹏谢佳成熊国顺

【Author】 Zhang Congpeng;Xie Jiacheng;Xiong Guoshun;School of Mechanical and Materials Engineering, North University of Technology;

【机构】 北方工业大学机械与材料工程学院

【摘要】 血细胞显微图像拼接时存在特征点冗余导致匹配效率低和过拟合的问题。基于自主搭建的血涂片自动扫描显微视觉平台,系统地研究高效、精确拼接技术。研究Forstner图像特征点检测的性能指标,发现Forstner算子具有较高准确度;针对Forstner算子检测特征点冗余的问题,提出区域匹配和图像定向滤波的复合预处理算法,减小无效搜索区域和图像噪声信息。实验结果表明:采用Forstner算子+复合预处理算法,图像拼接效果良好。对比传统全区域搜寻、匹配特征点的算法,匹配率提升58.69百分点,拼接耗时缩短了90.13%。

【Abstract】 There are problems such as redundant feature points due to matching inefficiency and over fitting during stitching when blood cell micrographs are stitched. Based on the self-built automatic scanning micro-vision platform for blood smears, efficient and accurate stitching technology was studied systematically. The Forstner operator performance was analyzed, and it was found that Forstner operator has higher accuracy. Aimed at the redundant feature points of Forstner operator, a composite preprocessing algorithm of region matching and image orientation filtering was proposed to reduce the invalid search area and image noise. The experimental results show that the image stitching effect is good while using the Forstner operator and the composite preprocessing algorithm. Compared with the traditional full area search and matching algorithm for feature points, the matching rate is increased by 58.69%, and the stitching time is reduced by 90.13%.

【基金】 北京市教委基本科研项目(110052371803)
  • 【文献出处】 计算机应用与软件 ,Computer Applications and Software , 编辑部邮箱 ,2022年09期
  • 【分类号】R446;TP391.41
  • 【下载频次】86
节点文献中: 

本文链接的文献网络图示:

本文的引文网络