3D Object Detection Using a New Descriptor with RGB-D

Abstract

Object detection is a very important study area in computer vision. Many research use only RGB images to find objects. In our work, we present new descriptor for object detection using RGB-D’s Depth image data. We combine RGB image with depth image to create new feature vector. The introduced features feeds Bag of Visual Words algorithm to classify images of the objects. Result shows us to RGB-D images are given better accuracy results to comparing with RGB image.