A comparative study of four enhancement algorithms traditionally used in computer vision for photometric normalization of images affected by illumination changes is presented in this paper. We experimented with the performance of these approaches to reduce the low frequency multiplicative noise that is produced as a result of a non-homogeneity illumination or a non-homogeneity developed chemical process in polyacrylamide gel electrophoresis images for the purpose of automatic classification of deoxyribo nucleic acid (DNA). The algorithms are tested on a database and their results are compared in a system for feature extraction and DNA classification.