Image retrieval by using content analysis is known as a difficult task. In our previous studies [1] and [2] mixed-metrics were proposed in order to combine color and texture metrics for image retrieval task. It was shown that it is always possible to mark out the best mixed-metrics for every group of similar images and improve retrieval effectiveness. In order to get the proper mixed-metrics a particular query-image should be classified to one of the predefined groups of perceptually similar images and this should be done in the real-time mode while processing the query and retrieving the result. In our previous work [3] the highly specialized classification method was proposed to solve this task. In the current study Naive Bayes and SVM classifiers are discussed and applied to the mixed-metrics approach to image retrieval. Classification result for these classifiers in comparison with the classifier, proposed in [3], are presented.