Description:
I need to capture as many images as possible continuously (viz. capturing 40 images per second) with higher frame rate and lossless compression but it should not be stored on any memory disk (because to save memory and to increase the processing performance ), as soon as it captures, immediately a specific portion of an image has to be analyzed whether it is bright or dark (say its capturing the Led light) and it should give output as 1 or 0 respectively.
Conditions:
1. I should be able to do programming only in C (if not possible then in C++) , I shouldn’t use any part of Java or Python programming (since timing matters a lot)
2. I don’t want to use OpenCV, Matlab or any other higher image processing softwares or tools (since requirement of my task is very simple and these Softwares might occupy higher ram,instead i could use this for higher processor perfromance)
CLARIFICATIONS:
I would be glad to receive your suggestions and clarifications regarding the following queries
1. Could you suggest me an open source platform (viz. softwares or tools) for performing the above mentioned task with the above specified requirements and conditions ?
2. I am not able to get a c programming code for real time continuous image acquisition using raspberrypi with higher frame rate, could you help me in providing the code or could you provide some valuable tips to accomplish this task.
Thanks in advance for your consideration and suggestions.
Hi,
If you want to do it in C/C++, the best way is to use OpenCV, it has a very good performances. You can find many simple code in the internet and adapt it to your use. I have worked a lot with OpenCV with C/C++ inetrface and it's not so hard and there is many documentation and example in the internet. Give it a try and you will see.
Regards,
Yacine YADDADEN.
If you need to access peripherals, the most efficient way is to use direct register access. Please see the following article:
http://hertaville.com/2014/07/07/rpimmapgpio/
If you have a video file stored somewhere, then I suggest you load a bunch of file data to ram and then perform the analysis. Otherwise, the computation can be very slow.
You can use CUDA if you have a GPU. This will allow you to process several images in parallel.
Hi,
Replica of c/c++ is inside Matlab. It is better to use Matlab. After loading the images in RAM from the devices by which you captured images then put all those images in a library function cell() which can store enormous images in the matrix form. The cell() function represents as an array of matrices. Use any loop of Matlab and perform the brightness and contrast test analysis one by one keep based on a criteria by taking sub-image window Sxy.
Hi everyone, thank you so much for your valuable answers, but here is my problem which is troubling me a lot, since requirement of my image processing operation is very simple, i dont want to use OpenCV or Matlab, could you suggest me any other light weight platform which i could perform the above operation ( as mentioned in my post) and most importantly it should be able to access the images from the ram and to carry out the processing operation and to discard it later after processing
- Badges
- Science topic