The Boeing Company

Hrl Laboratories, LLC

Theboeing Company

Terrell Nathan Mundhenk

Kyungnam Kim

Yuri Owechko

Heiko Hoffmann

A Arturo Flores

Michael J Daily

Changsoo Jeong

Howard Neely, Iii

Hai-Wen Chen

Deepak Khosla

Rajan Bhattacharyya

Kang-Yu Ni

David J Huber

Shinko Yuanhsien Cheng

Dmitriy V Korchev

Zhiqi Zhang

Arturo Flores

Described is a system and method for the classification of agents based on agent movement patterns. In operation, the system receives position data of a moving agent from a camera or sensor. Motion data of the moving agent is then extracted and used to generate a predicted future motion of the moving agent using a set of pre-calculated Echo State Networks (ESN). Each ESN represents an agent classification and generates a predicted future motion. A prediction error is generated for each ESN by comparing the predicted future motion for each ESN with actual motion data. Finally, the agent is classified based on the ESN having the smallest prediction error.

System and method for classifying agents based on agent movement patterns

A method includes receiving image data at a first tracking system. The image data represents a region in an image of a sequence of images. The method also includes generating a first tracking fingerprint based on the image data. The method further includes comparing the first tracking fingerprint and a second tracking fingerprint. The method also includes providing an output from the first tracking system to a second tracking system based on a result of the comparison of the first tracking fingerprint and the second tracking fingerprint. The output includes an instruction associated with an object model stored at the second tracking system.

System and method to improve object tracking using tracking fingerprints

Described is a training system for training a vision-based object detector. The system is configured to run an object detector on an image of a cleared scene to detect objects in the cleared scene. The object detector includes a support vector machine (SVM) or similar classifier with a feature model to generate an SVM score for object features and a spatial bias threshold to generate augmented object scores. The system designated detected objects in the cleared scene as false detections and, based on that, updates at least one of the feature model and spatial bias threshold to designate the false detections as background. The updated feature model or updated spatial bias threshold are then stored for use in object detection.

Training system for infield training of a vision-based object detector

A method includes receiving image data at a first tracking system. The image data may represent a region in an image of a sequence of images. The method includes generating a first tracking fingerprint based on the image data. The method includes comparing the first tracking fingerprint and a second tracking fingerprint. The method further includes providing an output from the first tracking system to a second tracking system based on a result of the comparison of the first tracking fingerprint and the second tracking fingerprint.

An apparatus, system, and method for increasing points in a point cloud. In one illustrative embodiment, a two-dimensional image of a scene and the point cloud of the scene are received. At least a portion of the points in the point cloud are mapped to the two-dimensional image to form transformed points. A fused data array is created using the two-dimensional image and the transformed points. New points for the point cloud are identified using the fused data array. The new points are added to the point cloud to form a new point cloud.

Backfilling points in a point cloud

Systems and methods are provided for backfilling a 3D cloud of coordinates. One embodiment is an apparatus that includes a camera able to capture an image, and a ranging system able to measure distances objects. The apparatus also includes a controller able to analyze the measured distances to generate a cloud of three dimensional coordinates, and to project coordinates of the cloud onto the image to match the coordinates with pixels of the image. The controller is also able to generate additional coordinates to increase the resolution of the cloud by iteratively defining a scanning window, detecting a group of coordinates that have been projected onto the scanning window, scoring each coordinate in the group based on its similarity to other coordinates in the group, dividing the scanning window into quadrants, selecting a coordinate from each quadrant, and generating an additional coordinate based on the selected coordinates.

Backfilling clouds of 3D coordinates

According to an embodiment, a method for generating a 3-D stereo structure comprises registering and rectifying a first image frame and a second image frame by local correction matching, extracting a first scan line from the first image frame, extracting a second scan line from the second image frame corresponding to the first scan line, calculating a pixel distance between the first scan line and the second scan line for each pixel for a plurality of pixel shifts, calculating a smoothed pixel distance for each pixel for the pixel shifts by filtering the pixel distance for each pixel over the pixel shifts, and determining a scaled height for each pixel of the first scan line, the scaled height comprising a pixel shift from among the pixel shifts corresponding to a minimal distance of the smoothed pixel distance for the pixel.

Stereo-motion method of three-dimensional (3-D) structure information extraction from a video for fusion with 3-D point cloud data

According to an embodiment, a 2-dimensional (2-D) image of a geographical region is transformed via a regional maxima transform (RMT) or an edge segmenting and boundary filling (ESBF) transform to produce a filtered 2-D image. The filtered 2-D image is iteratively eroded and opened to produce a processed EO 2-D image, 2-D object shape morphology is extracted from the processed EO 2-D image, and 2-D shape properties are extracted from the 2-D object shape morphology. A height slice of a 3-dimensional (3-D) point cloud comprising 3-D coordinate and intensity measurements of the geographical region is generated, and slice object shape morphology is extracted from the height slice. Slice shape properties from the slice object shape morphology are extracted, and the 2-D image is constellation matched to the height slice based on the 2-D shape properties and the slice shape properties.

Automated graph local constellation (GLC) method of correspondence search for registration of 2-D and 3-D data

A method of classification and segmentation of an image using modules on a computer system includes receiving a plurality of models having features suitable for classifying each pixel of the image into a respective one of a plurality of categories, using a classifier to provide a score for each pixel in the image for each category and using a segmenter to segment the image into image segments, wherein each image segment is a contiguous set of pixels having at least one common feature. For each image segment a set of average probabilities for each category is determined, and for each image segment, a most likely category to which the image segment belongs is determined by the maximum average probability resulting in a labeled segment image, which is used to identify any empty areas as incorrect holes. Then any empty areas that are identified as incorrect holes are filled.

Method for classification and segmentation and forming 3D models from images

The present invention relates to a system for detecting an object of interest in a scene. The system operates by receiving an image frame of a scene and extracting features from the image frame, the features being descriptors. The descriptors are quantized to generate PHOW features. A sliding window protocol is implemented to slide a window over the image and analyze the PHOW features that fall inside the window. Finally, the system determines if the PHOW features represent the object of interest and, if so, then designates the window as a location in the image with a detected object of interest.

System for detecting an object of interest in a scene

A method and system for processing a sequence of images using fingerprints. A current image in the sequence of images is segmented into a plurality of segments. A plurality of segment fingerprints is created for the plurality of segments. Segments in the plurality of segments are fused together using the plurality of segment fingerprints and a set of prior segment fingerprints created for a previous image in the sequence of images to form a set of master segments.

Method and system for processing a sequence of images using fingerprints

Methods and apparatus for use in reducing noise in one or more video frames. An expected value corresponding to each area (e.g., pixel) of a plurality of areas within the video frame is determined. A surprise value associated with each area is calculated based at least in part on the expected value and an actual value corresponding to the area. The surprise value represents a divergence of the actual value from the expected value. The response of one or more noise filters is attenuated with respect to a first area of the plurality of areas based on the surprise value associated with the first area.

Methods and systems for video noise reduction

Described is a system for scene change detection. The system receives an input image (current frame) from a video stream. The input image is color conditioned to generate a color conditioned image. A sliding window is used to segment the input image into a plurality boxes. Descriptors are extracted from each box of the color conditioned image. Thereafter, differences in the descriptors are identified between a current frame and past frames. The differences are attenuated to generate a descriptor attenuation factor &#x3b1;. Initial scores are generated for each box based on the descriptor attenuation factor &#x3b1;. The initial scores are filtered to generate a set of conspicuity scores for each box, the set of conspicuity scores being reflective of the conspicuity of each box in the image. Finally, the conspicuity scores are presented to the user or provided to other systems for further processing.

System and method for outdoor scene change detection

A method and system for tracking moving objects in a sequence of images. In one illustrative embodiment, a current image in the sequence of images is segmented into a plurality of segments. Segments in the plurality of segments belonging to a same motion profile are fused together to form a set of master segments. A set of target segments is identified from the set of master segments. The set of target segments represent a set of moving objects in the current image. A set of fingerprints is created for use in tracking the set of moving objects in a number of subsequent images in the sequence of images.

Growing community of inventors

Calabasas, CA, United States of America

Terrell Nathan Mundhenk