Simultaneous localization and mapping using multiple view feature descriptors

A system for determining a position of a camera using wide baseline matching, comprising:

• means for tracking within a sequence of images received from the camera having an incremental baseline change between two images, the sequence of images describing a three-dimensional environment surrounding the camera;
• means for receiving a recognition image from the camera and extracting a recognition feature descriptor from the recognition image;
• means for describing coupled to the means for tracking, the means for describing to generate one or more training feature descriptors using an approximate Kernel Principal Analysis (KPCA), a training feature descriptor based on the incremental baseline change between the two images in the sequence of images received from the camera and the training feature vector associated with position information describing a location of an object within the three dimensional environment relative to the camera;
• means for mapping a scene from the recognition image, coupled to the means for tracking and the means for matching, the means for mapping to create a three dimensional map of the three dimensional environment captured by the sequence of images using an extended Kalman Filter and the one or more training feature descriptors, said three dimensional map including position information associated with one or more feature vectors describing locations of one or more objects within the three dimensional environment relative to the camera;
• means for matching coupled to the means for tracking, the means for matching to match the recognition feature descriptor to a training feature descriptor from the one or more training feature descriptors and identifying a position information associated with the training feature descriptor from the one or more training feature descriptors which describes a location of an identified object within the three dimensional environment relative to the camera;
• and
• means for positioning, coupled to the means for mapping and the means for matching, the means for positioning to determine a position of the camera within the three dimensional map, and responsive to the appearance variation within the recognition image.

A computer program product, comprising a non-transitory computer-readable storage medium having computer program instructions and data embodied thereon for implementing a method for determining a position of a camera using wide baseline matching, the method comprising the steps of:

• receiving a sequence of images from the camera wherein there is an incremental change between two images, the sequence of images describing a three dimensional environment surrounding the camera;
• generating one or more training feature descriptors using an approximate Kernel Principal Analysis (KPCA), a training feature descriptor based on the incremental change between the two images in the sequence of images received from the camera and a training feature vector associated with position information describing a location of an object within the three dimensional environment relative to the camera;
• creating a three dimensional map of the three dimensional environment captured by the sequence of images using an extended Kalman Filter and the one or more training feature descriptors, the three dimensional map including position information associated with one or more training feature descriptors describing locations of one or more objects within the three dimensional environment relative to the camera;
• receiving a recognition image containing an appearance variation relative to at least a last image from the sequence of images;
• extracting a recognition feature descriptor from the recognition image;
• and
• determining a position of the camera within the map by matching the recognition feature descriptor to the training feature descriptor from the one or more training feature descriptors and identifying a position information associated with the training feature descriptor from the one or more training feature descriptors which describes a location of an identified object within the three dimensional environment relative to the camera.