Evaluación de la calidad de las imágenes de rostros utilizadas para la identificación de las personas

Automatic evaluation of face image quality is an important topic in the development of face recognition systems. This paper describes a framework for assessing the conformity of face images with the parameter set in the standard ISO/IEC 19794-5 that determines if given images possess an identification value and can be used in personal identification documents. New algorithms for image analysis and image classification with respect to parameters evaluated in the proposed framework are presented. The proposal is implemented as a dynamic link library (DLL), offering many advantages for the use in different applications. Good and bad quality images are evaluated, and the feasibility of the use of this framework is shown experimentally.