40 lines
1.6 KiB
Markdown
40 lines
1.6 KiB
Markdown
|
|
# LAB2 Project Presentation
|
|
|
|
## Project Description
|
|
|
|
The LAB2 project consists of the design and implementation of a digital system for image processing and data communication. The system is composed of several functional blocks, each responsible for a specific task in the processing pipeline. The main objectives are:
|
|
|
|
- Receive and process image data
|
|
- Perform color space conversion (e.g., RGB to grayscale)
|
|
- Apply convolution filters to the image
|
|
- Packetize the processed data for transmission
|
|
- Support loopback and test modes for verification
|
|
|
|
## Block Diagram
|
|
|
|
Below is a conceptual block diagram of the LAB2 system:
|
|
|
|
```mermaid
|
|
flowchart LR
|
|
IN[PC] -.-> UART[UART]
|
|
UART -.-> IN
|
|
UART --> DEPACK[Depack]
|
|
DEPACK --> C2G[Color to Grayscale Converter]
|
|
C2G --> BRAM[BRAM Writer]
|
|
BRAM --> Underflow
|
|
BRAM -->Overflow
|
|
BRAM -->Ok
|
|
BRAM <-.-> CONV[Convolution Filter]
|
|
BRAM <-.-> CONV
|
|
CONV --> PACK[Packetizer]
|
|
PACK --> UART
|
|
```
|
|
|
|
## Areas for Improvement
|
|
|
|
- In loopback mode, when sending an empty packet (FFF1), the module temporarily stores H and F and prepends them to the header of the next packet. There is an error in the pack/depack logic.
|
|
- The implementation of the depacketizer is complex, and the precise error has not been identified.
|
|
- The color conversion (C2G) uses a divider by 3, but the approximation method is unclear (it sums half of the power-of-two factor used in the divider—why?). -> Just add comments explainig how it works
|
|
- Convolution is performed with various unconstrained integers.
|
|
- In general, the VHDL code is somewhat complex, although generally correct. Aim to simplify and make the code more readable. |