Main Article Content

Authors

A method for calculating a Quality Index (Quality Index - QI) in images, which is the linear combination of three measures of perceptual distortion based on Human Visual System is presented. These three measures are projected to a multi - resolution space with operators in wavelet space, in which extraction of features such as homogeneity, contours and edges is carried out. To validate the method, the calculated QI is contrasted with an arithmetical mean observation score, which is determined experimentally on a standard set of images. The results show that the multi - resolution techniques allow a more appropriate description of the perceptual metric, reducing the computation time, as long as measurements are a priority over the details. Likewise, measurement processes are calculated on the same representation space, avoiding additional processing to extract features. This quality index is used in a face detection application to define the minimum distortion condition to classify images.

Benhur Ortíz Jaramillo

Universidad Nacional de Colombia Sede Manizales.
Grupo de Control y Procesamiento Digital de Señales

Julio César García Álvarez

Universidad Nacional de Colombia Sede Manizales.
Grupo de Control y Procesamiento Digital de Señales

Germán Castellanos Domínguez

Universidad Nacional de Colombia Sede Manizales.
Grupo de Control y Procesamiento Digital de Señales
1.
Ortíz Jaramillo B, García Álvarez JC, Castellanos Domínguez G. Translating distortion measures to an multi-resolution spaces for image analysis. inycomp [Internet]. 2010 Jun. 9 [cited 2024 Nov. 5];12(1):43-55. Available from: https://revistaingenieria.univalle.edu.co/index.php/ingenieria_y_competitividad/article/view/2701