Video Beginner Series 9: YUV/YCbCr Color Space on AXI4-Stream Interface

[florentw]

Introduction

This Video Beginner Series 9 is an introduction to YUV/YCbCr color spaces and how to convert video data from/to YUV/YCbCr on Xilinx devices.

---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

Summary

1. YUV and YCbCr color space

2. Converting YUV/YCbCr from/to RGB on Xilinx devices

3. Tutorial - Converting video data from YUV to RGB color space

4. What next?

[florentw]

YUV and YCbCr color space

What is a color space

A color space is a way to represent colors. In a computer you will often talk about RGB (for Red Green Blue). This is easy to visualize for human brain as it is like when you play with paint. If you mix some Red pain with some blue pain, you will get a Violet color (it is called additive colors).

This is the same for a pixel in the RGB color space. If its blue and red components have a high value while its green component is low, the pixel will be violet.

YUV color space

First televisions were in Black and White (B&W). When the first color television arrived it needed to be compatible with B&W TV infrastructure while being able to get color information. The B&W system used only the luminance information (represented Y). The color information (called chrominance and represented by U and V) was added on top of it.

YCbCr

YCbCr is a scaled and offset version of the YUV color space. Y also represent the luminance and Cb and Cr represent the choma components containing the color information.

There are different standard defining YCbCr as per example ITU-R BT601, ITU-R BT709 or ITU-R BT.2020

[florentw]

Converting YUV/YCbCr from/to RGB on Xilinx devices

Xilinx Video IPs to convert YUV/YCbCr from/to RGB

Xilinx IP catalog contains 3 IPs (dedicated) to convert YUV/YCbCr data from/to the RGB.

The recommendation is to use the Video Processing Subsystem (VPSS) IP. The documentation for the VPSS can be found in PG231. One thing to note is that the VPSS IP requires the use of a processor (Zynq or MicroBlaze) to be programmed through an AXI4-Lite.

For design without processor, a solution is to use the logiCORE RGB to YCrCb Color‐Space Converter IP (documented in PG013) or the logiCORE YCrCb to RGB Color‐Space Converter IP (documented in PG014).

YUV/YCbCr mapping on the AXI4-Stream Interface

On an AXI4-Stream Interface connecting a Xilinx Video IP (as defined by UG934), the Y component is mapped on the LSB, followed by the Cb component and then the Cr component

[florentw]

Tutorial - Converting video data from YUV to RGB color space

Note 1: This tutorial is intended to be used only with Vivado 2018.1 and only in simulation

Note 2: A valid license for the Test Pattern Generator is required to build the design.

Build the Vivado project

1. Download the tutorial files and unzip the folder
   
   
2. Open Vivado 2018.1
   
   
3. In the tcl console, cd into the unzipped directory (cd <path>/XVES_0009)
   
   
4. In the tcl console, source the script tcl (source ./create_proj.tcl)
   
   
5. Open the BD design. You can see that it only contains a test pattern generator controlled by a VIP IP.
   
   

Get the RGB output

6. Run the simulation for 6ms.
   
   
7. An output image proj_1/proj_1.sim/sim_1/behav/xsim/TPG_RGB.ppm was created. Open it with GIMP. We can see that this is the color bar pattern in TGB color space

Configure the TPG IP to output YUV data

8. Configure the TPG IP to output YUV. Open the test bench file tb_YUV_RGB.sv and find lines 184-185:
   
   

//Set TPG output color space
master_agent.AXI4LITE_WRITE_BURST(TPG_COLOR_FORMAT_REG,0,TPG_RGB_OUT,resp);

Change line 185 to:

master_agent.AXI4LITE_WRITE_BURST(TPG_COLOR_FORMAT_REG,0,TPG_YUV_OUT,resp);

9. Additionally, change the output file name to TPG_YUV.ppm on line 230
   
   
10. Save the test bench file and re-run the simulation for 6ms
    
    
11. If we open the new image file crated proj_1/proj_1.sim/sim_1/behav/xsim/TPG_YUV.ppm we can see that it looks like the same pattern but with different colors. This is because we are trying to display YUV colors while GIMP thinks it is RGB

12. A better way to look at it would be to look at each individual layer. In GIMP, click Colors > Components > Decompose
    
    
13. In the Decompose window, select RGB as Color Model, uncheck Decompose to layers and click ok

Note: The reason why we select RGB is because GIMP thinks our input is RGB. The Decompose will with RGB settings (if input is RGB) will only split the image into 3 different images (1 for each component).

14. 3 new windows will open, one for each component.

 

15. The window called YUV-red.ppm correspond to the Y component, the YUV-green.ppm and YUV-blue.ppm to Cb and Cr components.

Converting the AXI4-Stream YUV data to RGB

16. Open the block design.
    
    
17. Delete the wire between the AXI4_Stream_out output and the TPG
    
    
18. Right-click on the BD and click Add IP…. Search for ycrcb2rgb and press ENTER.

19. Double click on the YCrCb to RGB IP to open its configuration GUI. Change the following parameters:

• Pixels Per Scanline: 640
• Scanlines per Frame: 400
• Standard Selection: HD ITU 709 1250 PAL
• Output Range Selection: 0 to 255 for Computer Graphics

20. Connect the video_in input to the TPG m_axis_video output and the video_out to the BD AXI4_Stream_out output
    
    
21. Connect aclk and aresetn to the ports aclk_50MHz and aresetn_0

 

22. Validate and save the BD.
    
    
23. Open the test bench file and change the output file name to converted_RGB.ppm. Save the test bench file
    
    
24. Launch the simulation and run it for 6ms.
    
    
25. Open the generated image proj_1/proj_1.sim/sim_1/behav/xsim/converted_RGB.ppm. We can see that we have the pattern in RGB at the output

[florentw]

What Next?

• Do you have issues/questions following this Vivado Beginner Series?
  1. Search on the Xilinx forums for similar questions
  2. Create a new topic on the Video Board for your issue/question with the title starting with [Video Beginner Series 9] and followed by a quick description of your issue/question
     
     
• You think you master this subject?

• Look if there are unsolved topics on the Xilinx Forums about this Vivado Beginner Series you can answer

• You liked this Video Series?
  
  • You can give Kudos using the Kudos button 
  • Make sure you are following the Xilinx Video Series topic to be informed when an new topic is published (Go to the Xilinx Video Series topic > Options > Subscribe)