I'm working on manually analyzing a series of photographs of archaeological artifacts. Each photo contains a series of stone tool fragments laid out on a neutral background. I'm using Fiji image analysis software and I would like to create a tool that, in one action, will take a measurement of a linear dimension AND leave a mark on the image of the artifact that was just measured (so that I don't repeat measurements). I have little programming knowledge, so please explain in very simple, step by step terms if possible.
THANK YOU
Could you please upload a sample image and indicate which dimension you want to have measured.. Thanks!
Thanks for responding. Here's a sample image of the stuff I'm trying to measure. I'm just trying to get the longest dimension of each piece. Keep in mind that I have to do this manually (I've experimented with the automatic processing...it doesn't give me precise, repeatable counts and the time involved in threshholding the images makes it uneconomical). The task of measuring with the linear select tool, hitting ctrl-m to measure, switching to the brush, marking the piece to show it as "measured" , and then starting over is simple enough, but I'd really like to streamline it to one or two clicks since I have to do this thousands of times. Any help is enormously appreciated.
- It would greatly help if you could illuminate the scene straight from above, so that the shadows that are cast are minimal. Because a 'simple' program will assume the shadow belongs to the object. And it would also greatly help if objects don't touch each other and don't touch the white papers.
- You want the longest dimension. If expressed in number of pixelsizes, is that good enough, or does the program need to recognize and interpret the ruler?
- Are the horizontal and vertical pixelsizes equal?
- It is not clear to me how you want to link measurements with the objects that are being measured. A textfile with x and y pixel-indices of the object (e.g. center of mass), followed by it's length, would that be handy? Think about it...
From the ImageJ menu, select Analyze->Set Measurements...
On that dialog box, enable the checkbox for "Add to overlay"
Now follow your standard procedure:
1. Calibrate the image dimensions on your ruler.
2. Use the straight line tool to drag along the long axis of your stone tool.
3. Click Ctrl-M to record the length.
4. Repeat steps 2-3 for all stone tools.
With "Add to overlay" enabled, the yellow line segment will persist on your screen, and this will be your visual mark for each stone tool that has already been measured.
Cheers,
Mike
Agree with Lambert - the quality of the image isn't really that great, as there are (a) shadows and (b) specimens overlapping.
Don't know if you can recreate those images with less samples and a more uniform lightning. As the precision of the measurements is also determined by the numbers of pixels it might be a good idea to retake the photograph. This way you could use semi-automated thresholding, particle analyzer etc. But it really depends on (1) how many pictures you have to analyse and (2) how many objects you need to count.
Anyway, if you really want to try and get everything done manually, maybe make an adaption to your protocol:
- right click "duplicate" (just in case something goes wrong, you're not working on the original file)
- select linear Measure tool
- draw measure line
- go to "Analyze" "Label" (this labels the line)
- press Ctrl+M to measure (the measurement related to the labelled line is saved)
- copy/export the measurements to a *.csv file
- save the new image (with the labelled measurements)
You can also make your own Shortcuts (Plugins --> Shortcuts), so you can make this quicker (e.g. put label on F2, and measure on F3).
Hi,
I did linear measurements on about 10.000 platelets. I wrote a small macro to Image J, to make things easier for me.
After loading the macro into imageJ (or Fiji) my workflow is as follows:
1. Draw a line with the measurement tool
2. Press the button 1,2,3 or 4 on the number pad
The macro writes a number next to the measured structure to mark it as measured. Furthermore it adds the values to the ROI-Manager and the result table. The result table can than be saved as csv file and be imported into excel. The number 1,2,3 or 4 are used to mark different measurement types (i had length and width and some others).
If this would be of any use for you, I will happily share it to you.
You can find an example of the measured images with the assigned ROIs in Figure 3 and 4 in our publication linked below.
Best regards. Max
Article A New Method for Measuring the Dynamic Shape Change of Platelets
Hi,
if I understand you well, the 'Feret diameter' is Set Measrurements should be what you need. It measures the longest dimension of an object in any direction. The result for each recognized object goes to the Results table.
See http://imagej.nih.gov/ij/docs/guide/146-30.html#toc-Subsection-30.7
Best regards,
Jan
As far as I understand the application of Feret diameter would need the (automated) detection of the boundary of the object?
That's right, segmenation is necessary, it works on binary image. I didn't played with the provided image, but it seems on the first sight that converting to grayscale, background subtraction and threshold segmentation might by fine.
You are getting multiple answers about the picture quality.
If you are satisfied with the picture quality and are happy making thousands of measurements manually, then I wouldn't change anything.
If, however, you want to make tens of thousands of measurements automatically, then improving the picture quality such that you apply a single threshold value to the entire image and then automatically measure the Feret diameter of every object is the best way forward.
See if someone at the University has a camera stand with appropriate lighting. You may find this in the art department or in a natural history museum or in any lab that photographs specimens (such as yourself). Then the camera is mounted with the optical axis normal to the photographed surface and the shadows will be minimal.
Good luck.
- Badges
- Science topic
- Similar topics
- Engineering
- Electrical Engineering
- Signal Processing
- Image Processing