Welcome to bwtrack’s documentation!

bwtrack is a Python package for tracking black and white particles in the mini interfacial rheometer developed by my friend Qiaoge. See source code and notebooks on GitHub.

乔哥牛逼！

I’ve got very timely and insightful feedbacks from Qiaoge, and I’m optimistic that the method here will become useful for other purposes eventually. So I create this repo to record the progress of the core functions.

I’ve always been fascinated about how a software package is made available to users from general public. I have downloaded many Python packages from PyPI. It is part of my habit to check out if there are existing projects that aim to solve my problem, and I’m more than happy to go through the documentations and see if there is a chance to participate in the project.

I’ve also been interested in publishing the code I write for other people to use. This is the key motivation for me to write code in a maintainable way, with clear comments and documentations. I tried to convince other people to try my code, but every time I have a feeling that I’m confusing other people with some functions in my local library, which do not exist on their computer. The problem is, I import my own functions as if they are from available packages, but they are not. For example, I often write in my code

from myimagelib.myImageLib import readdata

and this almost always cause problem to other people. Because they will get an error “myImageLib not found”. Users with some Python experience will try

pip install myimagelib --upgrade

only to find there is no package called this. Then they come ask me, and I explain them: you need to go to my GitHub and download this piece of code to your computer, and set this directory as your Python path. Not many steps, but can be confusing at each step, and … not easy.

Imagine, if I can make myImageLib a PyPI package, and state it as a requirement of other code I give to users, things can be much more straightforward!

So this code, along with myImageLib, will be the first package I’m going to publish. The goal is to make it downloadable through pip, and work properly right after downloading.

Contents

• Usage
  • Installation
  • Minimal example
• API reference
• Notebooks
  • 尝试在乔哥的图片上找到黑白两色的颗粒
    • 1 用Python代码实现TrackMate的LoG方法（非完全复现，没有细节，只有基本思路）。
    • 2 区分黑白颗粒
    • 3 寻找白色颗粒
    • 4 全部流程整合
    • 5 黑色particle重合问题
    • 6 每个颗粒的Trajectory
    • 7 Subpixel accuracy
    • Appendix: attempt to remove background inhomogeneity, no success
  • Improve the accuracy and efficiency of the tracking
    • 0 Run original code
      • 0.1 Find bw on the large image
      • 0.2 Plot particle locations
      • 0.3 Take a small area for this test
    • 1 Improve accuracy
      • 1.1 Misclassification of black and white
        • 1.1.1 What’s wrong with mean intensity?
        • 1.1.2 Seek better criteria
          • Standard deviation
          • Other criteria
      • 1.2 White false positive
        • 1.2.1 Mean intensity
    • 2 Improve efficiency
      • 2.1 Which step is slow?
      • 2.2 New find_white
    • 3 Last remark
    • 4 Minimal example
  • White particles
    • 1 The old method
    • 2 A closer look at the white particle pixel intensity profiles
      • 2.1 Particles that are found by find_white()
      • 2.2 Particles that are not detected
    • 3 A better mask?
      • 3.1 Try different parameters of Mexican hat
      • 3.2 Gaussian mask
      • 3.3 Double well
    • 4 Try find_black
    • 5 Add more optional args to find_white()
    • 6 Minimal example
  • Examples of use