Completed Lbp Based Texture Analysis in Mammogram

Breast cancer is a frequent cancer diseases and it is the leading cause of cancer death among women in most of the occidental countries. Mammography is one among the key tool to identify the location and size of tumor in the breast. Texture analysis plays an important role in detecting the disease patterns in mammogram and to identify the masses as normal or abnormal. The local binary pattern descriptor provides an illumination invariant and rotation invariant approach for the texture analysis. However the LBP consider only the sign parameters. So it may lose some textural information. This can be overcome by considering the sign, magnitude and centre gray level values. Here a new approach for the Texture analysis in mammogram using completed LBP is presented. Although different methods have been proposed most of them suffer from large number of false positives. In contrast this method uses textural properties to reduce the number of false positives.