About This Research
What if WebAssembly is not truly faster than JS everywhere? This project tries to understand whereS and howS.
Why Image Processing?
Why not? Image processing is a commonly used in most popular applications today and it requires a lot of CPU power, which makes it a suitable choice for optimizing with WASM.
Measured Metrics
-
Execution Time: Total processing duration in
milliseconds - Cold start: Time taken for the first run, including any initialization overhead
-
Pixel Rate: Number of pixels processed per second (
MP/s) - Consistency: Variability in execution time across multiple runs
- Image Size Impact: How performance scales with different image sizes
Performance Benchmarks
Color Inversion
Lightweight operation - inverting RGB values pixel by pixel
Edge detection
Medium-intensity operation - detecting edges using the Sobel operator
Color Quantization (K-Means)
High-weight operation - reducing colors using K-Means clustering algorithm
Methodology
Each test runs multiple iterations (1-30 depending on image size) to ensure statistical reliability. Results show the median value to reduce the impact of outliers caused by browser garbage collection or system processes. Run counts and other variables related to the tests are