ZYBO HDMI OpenCV HW accelerator (FAST corner)
ฝัง
- เผยแพร่เมื่อ 3 ม.ค. 2017
- This project performs the FAST corner detection algorithm on the ZYBO board with GNU Linux, OpenCV version 2.4.5 and the HLS video library to take advantage of the hardware acceleration (SDSoC). The video is grabbed from the HDMI input and the results are displayed after computation on the VGA port of the board (realtime).
Project on : www.jonataubert.com - บันเทิง
![บุษบา - เมนทอล [Official MV]](http://i.ytimg.com/vi/IGnWPokSEis/mqdefault.jpg)
Hi Im working on xilinx FPGA by using SDSoc. Can you tell me how to link the opencv library? I tried(2.4.5) to link them but some function like cv.imshow still prompt out "undefined reference". I appreaciate your help.
Hi, did you try to link it using my tutorial? Which version of SDSoC are you using? jonataubert.com/tuto/use_opencv_sdsoc.php
@@jonataubert yes I followed your tutorial.
Version 2019.1
@@yuhengkit4427 I didn't try on any other version than 2015.4 and never used cv.imshow in that context. Do cv.write and cv.read work? I currently don't have any valid sdsoc license... I will try to get one. Please email me (see about section of the website) so that we can find how to make it working ;-)
Hello, i'm very interested in doing image processing using a zybo board, but I have a few questions that I Hope that you have already the answers for them.
1st question is, can we use a usb camera with the zybo to catch video frames ?
2nd what should I do, to gain time and reach the goal of doing image processing using the zybo board, can you give me some advices ?
Thank you :)
Hi,
cool that you would like to play with that nice device :)
1) Yes, to do this, use it on the ZYBO with GNU Linux + OpenCV. Then, find a way to store each incoming frame in a specific (fixed) memory region (by using a pointer). This memory region will be shared with the "FPGA" part of the ZYNQ.
2) As you have the address of each incoming frame you can implement an SDSoC function (which is both a DMA + HW accelerator that process your data). This function will have to read data of the image that is located in the input video frame buffer and then, after processing, will output the processed image in an output frame buffer.
This output frame buffer could be read by your software or even by another hardware peripheral, such as a DMA that could stream the data to a video pipeline (this is what is happening in the video here, the output frame buffer is continuously read by a DMA -> VGA pipeline).
Just one important point: don't forget that when using GNU Linux the addresses in the user space are virtual.
Good luck ;)
PS: there are certainly some other ways to do this, please people, feel free to share your advice/experiences.
jonataubert Thank you so much for your advices
It seems to be a hard task to do, I'm actually reading " The zynq Book " in order to get a good idea about FPGA zynq, after that I will boot the zybo with Linux, then I will check every step that you told me
Thank you for your informations my friend :)
Can I have your e mail to keep the contact with you ?
This is a really good book, I also started learning with this one and the application notes from Xilinx :)
Just go to my website and you will find my email in "about" section.
Cheers
Was there much of a latency between controller and screen, about the same without the zybo or more?
Nvm, saw you held controller up to the screen. my b
Hi sorry for my late reply, the latency is ~10 horizontal lines of pixels when using the HW accelerator.
Please continue working on tutorial.