Getting images along a long band rapidly


Hello everybody,

I would like to perform an image analysis of deposits on a band of 40meters*20centimers width. The goal
is to get 160 images of 25cm (without recovery) along the 40meters to perform the analysis.

The images would be taken by moving the band with a crank below a box containing the image pickup device.

One solution could be to adjust each time the band and to take one by one all the photographies. However i think that
something based on video could be more efficient. Maybe I could take a video of the band and find a way to split this video in 160 images. Have you got some advices concerning this second possibility?

I see two ways for splitting the video : based on the running time (but this needs a constant speed of the band, difficult to get this by hand), or by pattern recognition (for example red lines on the band each 25cm). After this i suppose that i shoud extract the images of each sub-video and perform a kind of mosaicing to get my images…

In short, the process involved by this second possibility are not clear to me : is that even possible quite easily? How ImageJ could be involved in that?

Thanks for your help


Hi @searcher

The second option you describe could be performed by opening the video file in ImageJ. Then potentially skipping some time frames by creating a substack (Image > Stacks > Tools > Make Substack…) since, due to a certain frame rate, you might have many images overlapping quite a lot. After reducing the overlapping parts and this reducing the individusl image number, you can save the images as an image sequence in a specied folder. You can then use this folder with the Grid/Collection Stitching plugin (under Plugins > Stitching) to combine them to a single image along your 40m distance.
Hope this helps to get an idea on how to continue.


Thank you biovoxxel !

I will start some trials based on what you explained. I have few more questions :

For the moment my idea is to use VLC to extract the image sequence and to find the appropriate “ratio of extraction” to get enough recovery with a minimum number of images (because of format compatibility problems with ImageJ)

do you know a way to split the merged image of 40m into images of 25cm?
Is there any tool that i could use in a macro?