简体   繁体   English

ImageJ:图像稳定器

[英]ImageJ: Image Stabilizer

Problem: Stabilize a stack of images 问题:稳定一堆图像

Question: How could I perform image stabilization without key-point detections? 问:如果没有关键点检测,如何执行图像稳定? Is there any way to do it? 有什么办法吗?

I tried this: http://www.cs.cmu.edu/~kangli/code/Image_Stabilizer.html in ImageJ and it does stabilize. 我尝试了这个:在ImageJ中http://www.cs.cmu.edu/~kangli/code/Image_Stabilizer.html ,它确实稳定。 The method used is - Lucas-Kanade. 使用的方法是-Lucas-Kanade。 I checked many sources and it all needs key-points (like corners, etc) to be defined. 我检查了很多源,所有这些都需要定义关键点(例如拐角等)。 [Source: https://link.springer.com/chapter/10.1007/978-3-642-25975-3_34] [来源: https//link.springer.com/chapter/10.1007/978-3-642-25975-3_34]

Are there any ideas how I could stabilize the stack of frames without key-points? 有什么想法可以在没有关键点的情况下稳定帧的堆栈吗?

Images are microscopic cell images (with cell-division) 图像是微观细胞图像(具有细胞分裂)

If you don't wan't to use keypoints you need to look at all the pixels in the image and use a measure that can compare the similarity of two images, and try to optimize this similarity. 如果您不想使用关键点,则需要查看图像中的所有像素,并使用可以比较两个图像相似度的度量,并尝试优化该相似度。

These measures could be 这些措施可能是

  • Cross Correlation 互相关
  • Mutual Information 相互信息

etc. 等等

ElastiX is a tool that supports these two measures and more, and you can allign two images at a time, so you will need to do it sequentially. ElastiX是一种工具,它支持这两种以及更多的措施,您可以一次分配两个图像,因此您需要按顺序执行。

However, you will probably get just as good (and faster) results using keypoints. 但是,使用关键点可能会获得同样(更快)的结果。

声明:本站的技术帖子网页,遵循CC BY-SA 4.0协议,如果您需要转载,请注明本站网址或者原文地址。任何问题请咨询:yoyou2525@163.com.

 
粤ICP备18138465号  © 2020-2024 STACKOOM.COM