Histogram of Oriented Gradients (HOG) in Computer Vision

The Histogram of Oriented Gradients (HOG) is a feature extraction algorithm used for object detection and recognition tasks. It was first introduced by Navneet Dalal and Bill Trigs in their CVPR paper "Histograms of Oriented Gradients for Human Detection". HOG is a global descriptor method that extracts neighborhood information, such as texture and edge structures, from an image and compresses it into a feature vector. The algorithm involves calculating the gradient image, histogram of the gradient, and normalization. The gradient image is calculated using the Sobel operator, which measures the difference in intensity at grayscale. The histogram of the gradient is created by combining the magnitude and orientation of each pixel into a histogram with 9 bins. Normalization is then applied to reduce the influence of lighting and contrast variations. The resulting feature vector, called the HOG descriptor, is used as input for classification algorithms like SVM. HOG is particularly effective for detecting shapes and outlines, such as the human form, and is widely used in computer vision applications. The algorithm can be easily implemented using OpenCV, which provides a built-in HOG descriptor and a pre-trained SVM model for detecting people.