This study develops a classification algorithm designed to handle interval data and to apply effectively in image processing. The proposed algorithm utilizes an innovative measure called overlap distance to assess the similarity between two intervals within multidimensional space. In addition, it integrates an improved method for determining prior probabilities by employing a fuzzy clustering technique. Furthermore, the study introduces a classification rule based on the quasi-Bayes method specifically tailored for interval data. According to this rule, an interval is assigned to a particular group if it holds the highest prior probability and the minimum distance to that group. The proposed algorithm is systematically presented in a step-by-step manner, elucidated by a numerical example, and executed using a well-established Matlab procedure. Another significant contribution of this study is its application to images, wherein texture features are extracted and represented as two-dimensional intervals. The effectiveness and superiority of the proposed algorithm are demonstrated through its application to various sets of medical images.