Depth kernel descriptors for object recognition

Consumer depth cameras, such as the Microsoft Kinect, are capable of providing frames of dense depth values at real time. One fundamental question in utilizing depth cameras is how to best extract features from depth frames. Motivated by local descriptors on images, in particular kernel descriptors, we develop a set of kernel features on depth images that model size, 3D shape, and depth edges in a single framework. Through extensive experiments on object recognition, we show that (1) our local features capture different aspects of cues from a depth frame/view that complement one another; (2) our kernel features significantly outperform traditional 3D features (e. g. Spin images); and (3) we significantly improve the capabilities of depth and RGB-D (color+depth) recognition, achieving 10–15% improvement in accuracy over the state of the art.

Authors

• Liefeng Bo
• Xiaofeng Ren
• Dieter Fox

Keywords

• Kernel
• Three dimensional displays
• Vectors
• Shape
• Object recognition
• Feature extraction
• Principal component analysis
• Kernel Descriptors
• Local Features
• Depth Images
• Depth Camera
• 3D Shape
• 3D Features
• Local Descriptors
• Kernel Feature
• Support Vector Machine
• Gaussian Kernel
• Vector-based
• Shape Features
• Point Cloud
• RGB Images
• Depth Map
• Image Patches
• Linear Classifier
• Linear Support Vector Machine
• 3D Point Cloud
• Local Kernel
• Kernel Principal Component Analysis
• Depth Features
• Category Recognition
• Dimensional Feature Vector
• Local Shape
• Finite Vector
• Pyramid Level
• Histogram Features
• Physical Size