LBP’s simplicity, discriminative power, and computational efficiency make it a popular choice in various computer vision applications.It was first described in 1994. LBP is a type of visual descriptor that helps classify textures in images.
It is a face recognition algorithm know for its performance and how it can recognize the face of a person from both the front and side face. These are the basic steps involved in the algorithm.
1. Local binary pattern (LBP)-supported image enhancement is applicable only to the grayscale image. So first convert the images to grayscale.
2. It is computed as a binary encoding of difference in pixel intensities in the local neighborhood. Here the image is converted to a 3×3 matrix format and compares the center cells to the outer cells.
3. For each pixel within a cell, compare it to its 8 neighboring pixels and if the value of the centercell image is greater than the value of the outercell image, then set the value as 1. This operator assigns 0 if the neighbor pixel is smaller than the center value.
4. Do the same process for all pixel values present in image.
5. Then by comparing the values with the original iamge, face recognition is done.