Simple signs existent in mammograms for diagnosing breast cancer are considered to be microcalcifications or MCs. Therefore, true detection of MCs is needed to minimize schedule diagnosis, efficient care and death rate due to breast cancer. A challenging task is to evaluate and interpret mammograms and, moreover to the poor contrast consistency of MCs relative to the remainder of the tissue, the precise identification of MCs, such as the minor size and random shape and size of the MC clusters, has several obstacles. These restrictions in the manual analysis of MCs increase the demand for an automated recognition system to help radiologists in mammogram analysis and it is important to design strength algorithm for this purpose. The goal of this paper is to present an efficient procedure that can be used to enhance images for extracting features to give excellent classification. The classifier senses which the region was normal, benign or malignant. The performance of KNN classifier with fuzzy histogram equalization using Otsu’s multi-threshold segmentation gives excellent results in detection and recognition in mammograms for breast cancer distinguished in image mammograms obtained from the hospital.