Image Filter with Performance Measurement

This project demonstrates a custom image filtering application in C. The code loads a grayscale PNG image, applies a convolutional filter, measures the filtering performance, and outputs the filtered image. Key metrics such as memory bandwidth and pixel rendering speed are displayed in real-time. The project leverages cycle counting, low-level memory manipulation, and the libpng library for image handling.

Features

• Convolutional Image Filtering: Applies a 3x3 kernel to enhance edges in grayscale images.
• Performance Metrics: Measures execution time, memory throughput, and rendering speed for benchmarking.
• File I/O with PNG: Utilizes libpng to read and write grayscale PNG files.
• Cycle-Based Timing: Uses processor cycle counting to calculate filter execution time.

Requirements

• C Compiler (e.g., gcc)
• libpng: PNG library for C (sudo apt-get install libpng-dev on Debian-based systems)
• GNU Make: For compiling the project

Installation

1. Clone the Repository:

   git clone https://github.com/ksemk/imageConvolution
   cd imageConvolution

2. Build the Project: Compile the program by running:

   make

   The makefile is configured to handle the compilation process automatically.

Usage

Run the executable with a grayscale PNG file as input:

./program input.png

This command performs the following steps:

• Loads the input image
• Applies the convolutional filter to enhance edges
• Measures performance metrics such as execution time, memory throughput, and rendering speed
• Outputs the processed image as out.png

Example Command

./program sample.png

Output

• Console Output: Displays performance data including cycle count, execution time, memory bandwidth, and rendering speed.
• Filtered Image: The resulting image is saved as out.png in the current directory.

Performance Measurement

The measure_performance function calculates metrics based on CPU cycles to assess filtering performance:

• Cycles: Number of CPU cycles used for filtering
• Execution Time: Time taken to apply the filter, calculated from cycles and CPU frequency
• Memory Bandwidth: Average data throughput in GB/s
• Rendering Speed: Pixel processing speed in Mpx/s

Code Overview

• filter: Applies a convolutional kernel to each pixel to highlight image edges.
• measure_performance: Measures performance metrics.
• main: Manages file I/O, initializes libpng, and coordinates the filtering and performance measurement functions.

Dependencies

This project uses the following files:

• program: Main executable file
• makefile: Automates compilation
• timestamp64.s: Assembly code to count CPU cycles
• LICENSE: License file for open-source usage

Contacts

For any inquiries, support requests, or if you're interested in collaborating, please don't hesitate to reach out:

• Email: kyrylo.semenchenko@gmail.com
• GitHub: Visit my GitHub profile
• LinkedIn: Connect with me on LinkedIn

I look forward to hearing from you!

License

This project is licensed under the MIT License - see the LICENSE file for details.