VapourSynth PSI Plugin

VapourSynth plugin implementing the PSI (Perceptual Sharpness Index) algorithm for measuring image sharpness perception.

Overview

The PSI metric provides a perceptual measure of image sharpness based on local edge gradient analysis. It evaluates the perceptual quality of edges in images by analyzing edge width characteristics across different image blocks.

Usage

PSI

Calculates the Perceptual Sharpness Index of input frames:

core.psi.PSI(clip clip[, float percentile=22.0, int blocksize=32, float threshold_w=2.0, float angle_tolerance=8.0, float w_jnb=3.0, float sobel_threshold=0.1, int output_mode=0])

Parameters:

• clip: Input clip (grayscale or YUV - only luma plane is processed) (format: 8-16 bit integer or 32 bit float)
• percentile: Percentage of sharpest blocks to use for metric calculation (range: 0-100, default: 22.0)
• blocksize: Size of blocks for averaging edge width measurements (default: 32)
• threshold_w: Minimum sum of edge widths in a block to process it further (default: 2.0)
• angle_tolerance: Tolerance for horizontal edge detection in degrees (range: 0-90, default: 8.0)
• w_jnb: Width threshold for Just Noticeable Blur correction (default: 3.0)
• sobel_threshold: Threshold for Sobel edge detection (default: 0.1)
• output_mode: Output mode selection (default: 0)
  • 0: Return original input frame (copy)
  • 1: Return sharpness distribution map (32-bit float grayscale)

Frame Properties

The filter outputs the following frame property:

• PSI: Perceptual Sharpness Index score (higher values indicate sharper images)

Building

meson setup build
ninja -C build install

Algorithm Reference

Based on the research paper:

• C. Feichtenhofer, H. Fassold, P. Schallauer
  "A perceptual image sharpness metric based on local edge gradient analysis"
  IEEE Signal Processing Letters, 20 (4), 379-382, 2013

and github repository: https://github.com/feichtenhofer/PSI