Car Detecting Method using high Resolution images

Abstract

A car detection method is implemented using high resolution images, because these images gives high level of object details in the image as compare with satellite images. There are two feature extraction algorithms are used for implementation such as SIFT (Scale Invariant Feature Transform) and HOG (Histogram of Oriented Gradient). SIFT keypoints of objects are first extracted from a set of reference images and stored in a database. HOG descriptors are feature descriptors used in image processing for the purpose of object detection. The HOG technique counts occurrences of gradient orientation in localized portions of an image. The HOG algorithm used for extracting HOG features. These HOG features will be used for classification and object recognition. The classification process is performed using SVM (Support Vector Machine) classifier. The SVM builds a model with a training set that is presented to it and assigns test samples based on the model. Finally get the SIFT results and HOG results, then compare both results to check better accuracy performance. The proposed method detects the number of cars more accurately