Local Binary Pattern Segmentation Algorithm



Local Binary Pattern (LBP) [3], [4], which has been used successfully to identify textures in faces and even in some EM images. To calculate LBP, a set of 8 pixels around a reference pixel is chosen, and the grayscale intensity of these pixels is compared to the reference pixel. The intensity of each of these 8 pixels can either be higher or lower than the reference pixel’s intensity, leading to an 8-bit “local binary pattern” associated with each pixel in the image. This 8-bit number for each pixel can be mapped to a new grayscale image, where now each pixel’s intensity value encodes some information about the spatial variation in grayscales in the original image. Hopefully this new LBP image will represent texture more tangibly than the original image. By only looking at relative intensity, LBP also has the advantage of being invariant to changes in global intensity, e.g. due to illumination differences.




[3] T. Ojala, M. Pietik¨ainen, and T. M¨aenp¨a¨a, “Multiresolution gray-scale and rotation invariant texture classification with local binary patterns,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 24, no. 7, pp. 971–987, 2002.
[4] S. Liao, M. W. K. Law, and A. C. S. Chung, “Dominant local binary patterns for texture classification.” IEEE Transactions on Image Processing, vol. 18, no. 5, pp. 1107–1118, 2009.

MATLAB code for converting a picture of a document into a binary image which fed to the Tesseract OCR

Convert ascii text to a binary image



There are no reviews yet.

Be the first to review “Local Binary Pattern Segmentation Algorithm”

Your email address will not be published. Required fields are marked *