ESIM
ESIM is an official MATLAB implementation of the Edge Similarity method for full-reference quality assessment of screen content images, published in IEEE Transactions on Image Processing. The method targets the human visual system's sensitivity to edges in screen content, extracting three salient edge features for comparison: edge contrast and edge width derived from a parametric edge model, plus edge direction computed directly from the image. These three similarity measures are combined through an edge-width pooling strategy to produce the final ESIM score. The package also includes or references the SCID database of 1800 distorted screen content images generated from 40 references across nine distortion types at five degradation levels. The implementation is intended for researchers working on image quality assessment, screen content coding, compression evaluation, and perceptual quality benchmarking. The code is released under the MIT License and provides all functions needed to compute ESIM scores given