University project for the Logical Networks course, Computer Science Engineering branch at Politecnico di Milano, A.Y. 2020/21.
The full specification can be found here (italian only) and a brief introduction can be found in the Specifications section below.
Devs: Javin Barone, Ottavia Belotti
The project's aim is to experience a low-level programming style simulating the programming of a FPGA. The challenge is to write very efficient code taking advantage (but at the same time dealing with the limitations) of the hardware architecture, rather than using high-level operations non-natively supported by the hardware. In fact, the project has to work with a clock period of at least 100 ns.
The project consists in a VHDL (VHSIC Hardware Description Language) implementation of a simplified image equalization algorithm. It supports 8-bit grayscale images and the maximum size for an image is 128x128 px.
An image is fed to the algorithm by storing sequentially each pixel value in the program memory starting from address 2, while cells 0 and 1 are reserved for the image size which has to be specified. Each pixel value (current_pixel_value) is analyzed and its new value is computed as follow:
delta_value = max_pixel_value - min_pixel_value
shift_level = (8 - floor(log2(delta_value + 1))
temp_pixel = (current_pixel_value - min_pixel_value) << shift_level
new_pixel_value = min(255, temp_pixel)
The programm processes all the values and writes the equalized version of the image right after the original image in memory, in a sequential order as well.
- XILINX Vivado WebPack (with target FPGA: xc7a200tfbg484-1)