sreevenkjan
Full Member level 5
Hi all,
I am designing a filter to do convolution of a binary image which is stored in the bram as a .coe file.I am able to do convolution and my logic is working fine.Well since images are 2d I need to do filtering in the horizontal direction(x direction) and also in the vertical direction(y direction).if I store in the block ram the image pixels in the x direction(i.e image row pixels) as 9 bits wide upto a depth of 25000 then my filtering logic in done only in the x-direction and if i store the .coe file with vertical pixel values(column pixels) then filtering is done only in the y direction and my output is not the correct one because i need filtering in both the direction....if i do both the filtering processes separately(storing 2 different .coe files of data) and then take the simulated data obtained and combine it...then my output is correct.
i have explained the problem above..so now my question is do i need to store two sets of data in block ram and then do the convolution or is there any other way to proceed.
Thanks,
Sreeni.
I am designing a filter to do convolution of a binary image which is stored in the bram as a .coe file.I am able to do convolution and my logic is working fine.Well since images are 2d I need to do filtering in the horizontal direction(x direction) and also in the vertical direction(y direction).if I store in the block ram the image pixels in the x direction(i.e image row pixels) as 9 bits wide upto a depth of 25000 then my filtering logic in done only in the x-direction and if i store the .coe file with vertical pixel values(column pixels) then filtering is done only in the y direction and my output is not the correct one because i need filtering in both the direction....if i do both the filtering processes separately(storing 2 different .coe files of data) and then take the simulated data obtained and combine it...then my output is correct.
i have explained the problem above..so now my question is do i need to store two sets of data in block ram and then do the convolution or is there any other way to proceed.
Thanks,
Sreeni.