How to read pixel values from an image and give them to DCT core?

[guduru]

Hi,
As a part of my project iam working with DCT core(block size is 8*8) available in Xilinx core generator.For dct core i need to give the pixel values sequentially.Can any body please help me how to read the values from an image file and giving those values to the dct core...Any material is available regarding the above please let meknow..

 

[calm]

Using DCT core

you can design a interface to ram, use any ram-read tool to check it.

 

[guduru]

Re: Using DCT core

i cant understand can you please explain in detail

 

[amitgangwar_vlsi]

Re: Using DCT core

for converting image to the numeric value you have to use MATLAB.by using MATLAB you can do it.
first convert the image in matrix form.
then convert it in indexed image in which values can vary from 0 to 127.
your image would be of greater size.so you can partitioned it into blocks of size 8X8.then you can use 1 block at a time.then convert each value in the binary form.
now you can give these values(8 bits) rowwise on every clk pulse to the DCT block.
after 94 clks you will get the output columnwise serially.

if you further have any problem then ask me freely.

GOOD LUCK

 

[guduru]

Re: Using DCT core

Hi amit,
Thanks for your reply.I have done what you have said.using matlab i got the pixel values as binary values and i stored the values in a memory and applied it to the DCT core but when i applied the values to dct core available in the core generator iam not getting the output.In the data sheet i have seen FIFO module before the 2D DCT core.Do i need to built this FIFO or is it available within 2D dct core?can you please help me in this matter?
I have one more question i have to implement the 2D dct on the virtexIIpro FPGA.Can u please help me how to implement memory on hardware and storing the pixel values in the memory and giving those values to the 2Ddct module on the FPGA kit .please reply as early as possible....

Thank you.

 

[amitgangwar_vlsi]

Re: Using DCT core

hi....
tell me the process in detail...what you are doing....
i have already use the 2d dct core of xilinx and get output successfully....

just give the input till 64 clks and after further 30 clks you will get the output.
you have to do nothing with the code....ok

if you want to store image data on fpga.then you have two option which i know...
1] write such type of code that instantiate Block RAM on FPGA(see help in xilinx).and store one pixel value at a time to the RAM.in this case you have to wait for 64 clk to store data in memory and then 64 clk to give input to the dct code and further 30 clks for output.

2]you can instantiate block ram on fpga(see language tempelate in xilinx).
and initilize that ram with the pixel values.in this case you required only 94 clks for getting output.

ok

if you have further problem, send me ur code i will check it...

amit gangwar

 

[guduru]

Re: Using DCT core

Hi,
DCT is working properly but the output values are not correct.i have sent you vhdl code any changes i have to make please let me know.
thank u

 

[amitgangwar_vlsi]

Re: Using DCT core

when you convert the image into the matrix in matlab....
at that time change the bwlabel of image in the matlab and do vary the pixel values in the range 0 to 127.....try it

 

[deh_fuhrer]

Re: Using DCT core

use the Matlab's command , "Imread" for this..it will give you the pixel values....If the image is colour one then u have to extract the three layers n have to process these individually (i m not really sure abt as i hav worked on dis in my bachelors project n have to recall it...) anyways if u face any problem feel free to contact me .....

Thanks,
:|

 

[WAJDI]

Re: Using DCT core

hiiiiiii
please i have the same problem with using 2D-dct from xilinx core generator
to implement it in viretx 2pro ,with powerpc processor
so ,i used two ram from xcore generator
my architecture to load the pixels of image into the ram and then the 2D-dct extract the bloc 8x8 pixels one after one in the right order,
i havn't succeed to write a code vhdl how control this operation of extracion
my question is ,in the ram i load the pixel line per line both the image source
or i can load this pixels in order block per bloc

please help me.

thanks for all******

 

[deepu_s_s]

Using DCT core

Hi guduru,
Did u implemetned 2D-DCT in FPGA. If so do u have any docs ?

Thanks and Regards
Deepak