Colorectal cancer is the fourth leading cause of cancer-related deaths in Iran. Nearly 50 % of patients diagnosed with colorectal cancer will eventually die from this disease. Since this cancer mostly does not have any sign, it is usually ignored by patients. Furthermore, over 40% of gastric malignancies appear as polyps. Early stages of colon cancer start with the emergence of small bumps called polyps. Although polyps generally do not have malignant cells, with the passage of time and the growth of polyps, they may become cancerous. That is why early detection of polyps, and removing them effectively reduce the number of people with colon cancer. The procedure of screening colorectal disease generally includes a fecal occult blood test, flexible sigmoidoscopy and colonoscopy; the last one is the most common non-invasive method. In this method, if a gland is observed, its tissue samples can be removed by colonoscopy. Research shows that 25% of polyps in the colon are not recognized in the process of colonoscopy. Specialists manually search colonoscopy images in order to detect polyps. In this detection method, there is the possibility of misinterpretation due to the time-consuming operation and the structure of the colon that has many folds and movements. Therefore, in some cases, specialists cannot detect small or even large polyps. By studying the polyps within the colon in terms of location, size, shape and texture, polyp features can be identified; and by using these features, we can design image processing algorithms for the automatic detection of polyps on suspicious images. These algorithms can be used to help specialists to identify areas where polyps are detected so that they can enhance the accuracy of diagnosis. Research in the field of polyp detection in colonoscopy images is mainly divided into two general methods: texture-based methods and shape-based methods.This project proposes a shaped-based method for detecting polyps. The algorithm used in This project is generally divided into three stages. The first one extracts image edges by Susan edge detector. In the second one, Niblack binarization is used to convert gray-scale image to binary one, and in the last stage,a new algorithm based on Hough transformation is used to locate polyps in the image. Keywords Colonoscopy, polyp, Colon cancer, Hough transform, Niblack binarization, Susan edge detector