Divyansh Manocha

Overview

This was a first year project at Imperial College London in a team of three: Parallel Image Processing using FPGA’s. Responsible for the reading and alignment of the live video feedback; added the ability for a user to control tempo

Project Specification

The task assigned was to implement parallel image processing using a FPGA. Considering the duration and resources provided for the project, it has been decided that a music playback system which processes musical notation will be implemented.

Basic image processing techniques were first implemented, including:

• Scheduling
• Blur filters
• Position tracking
• Edge detection (sobel filters)

Approach

Details from the image processing exercises as well as further concepts, such as line detection using the Hough transform (Hamarneh, Althoff and Abu-Gharbieh, 1999), will be used to achieve this task.

A flowchart describing the function of the system In order to complete the task in the given time the system shall use its own musical notation so a simpler algorithm can interpret each note. The initial approach to interpret modern music notation presented a challenge that could not have been overcome within the time constaints imposed. Instead, the pitch, duration and order will be determined from a custom grid notation.

Project Outcomes

Successfully implemented a live sheet music reader using Catapul C.

Supported Features

• Producing an appropriate notation
• Grid Detection
• Hough Transform Matrix
• Pixel and Square detection
• Tone Production Future work may involve the implementation of arrays, pointers and floating point arithmetic.