doiriel
commented
6 years ago Local Binary Patterns Applied to Breast Cancer Classification in Mammographies Operator to extract texture features Very popular for face detection and recognition Robust to illumination variations It is possible to obtain rotation invariance by grouping the extracted features in histograms Main parameters: number of neighbors, radius of comparison Histogram of binary values: 1 if a neighbor is greater than the central value, 0 otherwise False Positive Reduction in Mammographic Mass Detection Using Local Binary Patterns LBP to characterize micropatterns (i.e. edges, lines, spots, flat areas) and preserve at the same time the spatial structure of the mass Construction: 1 if a neighbor is greater than the central value, 0 otherwise. When all of the pixels have been labeled the histogram of labels is computed and used as a texture descriptor. ROI LBP Classifier ROI Segmentation using Local Binary Image Central pixel value in 3x3 block is thresholded on the basis of gray level values in its 8-neighborhood. The central pixel will be assigned 1 if any of the pixel in the 8-neighborhood does not equals the intensity of the central pixel. Otherwise intensity of central pixel is set to 0. In this way threshold T is calculated automatically for the targeted mammogram patch. To extract the mask image the double eroded image is converted into a binary image using the threshold calculated in the previous step. For the binary image obtained, connected components in the image are identified as regions and the region with maximum area is considered which gives final mask image.
Example with open CV https://www.pyimagesearch.com/2015/12/07/local-binary-patterns-with-python-opencv/
How we can use this method for mass detection?
@doiriel I have closed issue #51, so I created new issue for the investigation that you have been planning to do on Local Binary Patterns.