作者：禅与计算机程序设计艺术
1.简介
Local Binary Pattern (LBP) is a feature descriptor for texture classification and object recognition [1]. It was introduced by Leibe et al in their paper "Robust textural features for image classification" in 2002. The LBP algorithm has been widely used as an efficient alternative to convolutional neural networks for computer vision tasks such as object detection or face recognition.
The basic idea of the LBP algorithm is based on calculating the difference between adjacent pixel intensities. By considering only these differences, we can identify regions with different texture characteristics even when they are rotated or distorted. In other words, it extracts simple but effective binary patterns that contain information about the edges and corners of objects.
One drawback of traditional edge detection methods like Canny edge detector and Sobel filter is that they detect too many false positives due to e