Image Processing in FPGA

[fradaric]

Dear all,
I did canny edge detection in matlab, now I want to do the same in FPGA. Please tell me how can I load image file into FPGA. Here is my doubt,

1) The image file size is 256*256, but when I convert it into .coe/text format, image becomes 8*65536 size, how can I analyze it??? All my analysis is based on pixel values(gray scale)!
2) What are the possible ways of doing this in FPGA?, I can use any FPGA boards(Spartan, vertex,altera etc.)

Thanks in Advance!!

F JOe :wink:

 

[TrickyDicky]

256x256 is 65536. You know its 256x256, so you just keep track of where you are. How you analyse it is up to you.

there are lots of ways. Image processing is often done in fpga. I suggest finding a good book ln digital design.

 

[fradaric]

256x256 is 65536. You know its 256x256, so you just keep track of where you are. How you analyse it is up to you.

there are lots of ways. Image processing is often done in fpga. I suggest finding a good book ln digital design.

Thank you !!

 

[amitjagtap]

you can use MATLAB to read an image in ‘jpg’ format and to convert it to txt file. read image using txt format. try to keep the image format in 256*256 size. you can use virtex 4/5/6 FPGA and Spartan-6 and above... Spartan 3 FPGA may give resource crunches if low logic device is selected...

 

[fradaric]

you can use MATLAB to read an image in ‘jpg’ format and to convert it to txt file. read image using txt format. try to keep the image format in 256*256 size. you can use virtex 4/5/6 FPGA and Spartan-6 and above... Spartan 3 FPGA may give resource crunches if low logic device is selected...

Thank you, let me try with that, I will keep update both progress and issues ...

 

[syedshan]

256x256 is 65536. You know its 256x256, so you just keep track of where you are. How you analyse it is up to you.

there are lots of ways. Image processing is often done in fpga. I suggest finding a good book ln digital design.

Hi
do we not need any external memory here. Is the fpga it self ok and enough for all this?

 

[TrickyDicky]

It depends on the fpga.

 

[mano1988]

I believe you could try with System Generator tool from xilinx to perform DSP operations on xilinx FPGA. The image can be easily loaded into the FPGA through few simulink blocks. The HW-Co simulation can be performed over the FPGA through the tool.

 

[Chan Dev]

All you need is to create a ROM block to store a image in FPGA so that you could process..
there are certain matlab codes available online to convert image to mif format so that you can store it in ROM block and process it..

 

[shatruddha]

Well Its possible to use ROM block, but I would suggest to use an external memory. That way you will have your FPGA memory reserved for the processed data and run time data. I'm not sure how much data memory you are using currently, but as the Processing gets complex, data storage requirement boosts

 

[Chan Dev]

When you go for off chip memory how could you access the data...need of driver isn't complex..???

 

[TrickyDicky]

When you go for off chip memory how could you access the data...need of driver isn't complex..???

You need a memory controller - which are readily available or you could write your own

 

[shatruddha]

I dont think that will be much of a problem here. External memories are available with SPI communication, So using a SPI will not be a problem for you

 

[Donna111]

HI there
If you just want to load the image ,you can refer to this code:
namespace RE__Test
{
public partial class Form1 : Form
{
public Form1()
{
InitializeComponent();
}

private void button1_Click(object sender, EventArgs e)
{
string fileName = "c:/Sample.png";

REImage reImage = REFile.OpenImageFile(fileName);

REFile.SaveImageFile(reImage, "c:/reimage.png", new PNGEncoder());
}
It is from an image processing tool.Hope to help you.