Breast carcinoma is one of the most commonly occurring problem among women, which even leads to the end of the victim. Sonogram is widely used in imaging breast abnormalities. The paper is mainly concentrated in detecting lesion in breast sonogram images. The proposed scheme automatically identify the abnormalities without manual intervention. The ultrasound image is preprocessed by equalizing the histogram of the image followed by Gaussian filtering. The filtered image is then segmented using watershed technique which identifies the abnormal region both benign and malignant.