Métricas estructurales para evaluar la similitud de texturas naturales

This article evaluates the similarity of images of natural textures with changes of scale and rotation, through the application of structural metrics. The applied metrics are: Complex-Wavelet Structural Similarity Metric (CW-SSIM) and Structural Texture Similarity Metric (STSIM). First a decomposition of the images in sub-bands was realized. With the generated information the descriptors were extracted and the comparison formulas were applied and the strategies of union to obtain a value of total similarity. In experiments with texture images with rotation changes, the CW-SSIM and STSIM metrics showed average values of: 0.4857 and 0.9735 respectively. With respect to texture image tests with scale changes, the CW-SSIM and STSIM metrics showed average values of: 0.7002