Mammography analysis is a crucial tool when it comes to detecting breast cancer early. While analyzing a mammogram, the most crucial stage is the pre-processing because of the low quality of the original picture recorded. In order to fix and alter the mammography picture, pre-processing is crucial. A wide variety of pre-processing methods are at your disposal. Images taken from a Mammogram may include noise due to fluctuations in lighting and sensor inaccuracy. If these sounds can be eliminated without compromising the image's borders and small characteristics, a proper diagnosis of breast cancer may be made using imaging technology. In this study, we present a High-Density Noise Filtering (HDNF) technique for denoising digital mammography pictures. Experiments on a dataset including a wide range of mammography pictures test the effectiveness of the proposed technique against a variety of image quality measures. When the pictures have been denoised, a Region of Interest (ROI) technique based on statistical moments is proposed for locating suspicious areas in breast cancer images of varying modalities. When a ROI has been identified, its borders in the resulting subtracted picture must be defined. A Canny edge detection technique is used for this purpose. Mean square error, peak signal-to-noise ratio, and the Structural Similarity Index Method (SSIM) are used to evaluate the effectiveness of the Wiener Filter (WF), Gaussian Filter (GF), the Adaptive Median Filter (AMF), and the Hidden Markov Model (HDNF). Results from the experiment are presented in the form of a quality-of-images-measures comparison between the original, unscathed pictures and denoised versions of the same images affected by varied levels of generated salt-and-paper-noise and speckle-noise. The HDNF technique produces a higher quality result than the related filter strategy.

Mammogram, Denoising, ROI, WF, GF, AMF, HDNF, Canny edge detection.