RGB-D-Based Features for Recognition of Textureless Objects

  title={RGB-D-Based Features for Recognition of Textureless Objects},
  author={Santosh Thoduka and Stepan Pazekha and Alexander Moriarty and Gerhard K. Kraetzschmar},
Autonomous industrial robots need to recognize objects robustly in cluttered environments. The use of RGB-D cameras has progressed research in 3D object recognition, but it is still a challenge for textureless objects. We propose a set of features, including the bounding box, mean circle fit and radial density distribution, that describe the size, shape and colour of objects. The features are extracted from point clouds of a set of objects and used to train an SVM classifier. Various… 
1 Citations
b-it-bots: Our Approach for Autonomous Robotics in Industrial Environments
The b-it-bots team's combined 2D and 3D approach for object recognition has improved robustness and performance compared to previous years, and the task planning framework has moved away from large state machines for high-level control.


Ensemble of shape functions for 3D object classification
The presented shape descriptor shows that the combination of angle, point-distance and area shape functions gives a significant boost in recognition rate against the baseline descriptor and outperforms the state-of-the-art descriptors in the experimental evaluation on a publicly available dataset of real-world objects in table scene contexts with up to 200 categories.
Multi-view object recognition using view-point invariant shape relations and appearance information
An object recognition system coding shape by view-point invariant geometric relations and appearance is presented, which indicates that this approach can be used in contexts such as assembly manipulation which require high reliability of object recognition.
Multimodal templates for real-time detection of texture-less objects in heavily cluttered scenes
This work presents a method for detecting 3D objects using multi-modalities based on an efficient representation of templates that capture the different modalities, and shows in many experiments on commodity hardware that it significantly outperforms state-of-the-art methods on single modalities.
Multi-class fruit classification using RGB-D data for indoor robots
An effective and robust system to classify fruits under varying pose and lighting conditions tailored for an object recognition system on a mobile platform using RGB as well as depth cues for the specific technical setup of the robot.
Fast object recognition and 6D pose estimation using viewpoint oriented color-shape histogram
An object recognition and pose estimation framework consisting of a novel global object descriptor, so called Viewpoint oriented Color-Shape Histogram (VCSH), which combines object's color and shape information, which is proven to outperform other existing methods.
A combined texture-shape descriptor for enhanced 3D feature matching
The proposed descriptor, dubbed CSHOT, is demonstrated to notably improve the accuracy of feature matching in challenging object recognition scenarios characterized by the presence of clutter and occlusions.
Object discovery in 3D scenes via shape analysis
We present a method for discovering object models from 3D meshes of indoor environments. Our algorithm first decomposes the scene into a set of candidate mesh segments and then ranks each segment
Unique Signatures of Histograms for Local Surface Description
A novel comprehensive proposal for surface representation is formulated, which encompasses a new unique and repeatable local reference frame as well as a new 3D descriptor.
3D is here: Point Cloud Library (PCL)
  • R. Rusu, S. Cousins
  • Engineering, Computer Science
    2011 IEEE International Conference on Robotics and Automation
  • 2011
PCL (Point Cloud Library) is presented, an advanced and extensive approach to the subject of 3D perception that contains state-of-the art algorithms for: filtering, feature estimation, surface reconstruction, registration, model fitting and segmentation.
Visual pattern recognition by moment invariants
  • M. Hu
  • Mathematics, Computer Science
    IRE Trans. Inf. Theory
  • 1962
It is shown that recognition of geometrical patterns and alphabetical characters independently of position, size and orientation can be accomplished and it is indicated that generalization is possible to include invariance with parallel projection.