T K Holdings, Inc.

Automotive Systems Laboratory, Inc.

Joyson Safety Systems Acquisition LLC

Trw Automotive U.S. LLC

Gregory Gerhard Schamp

Owen A Davies

James C Demro

Michael E O'Boyle

Debbie E Nachtegall

William P Wernette

A structured light pattern is projected onto the path of a vehicle so as to generate a plurality of light spots, and an image thereof is captured from the vehicle. A world-space elevation of at least a portion of the light spots is responsive to a pitch angle of the vehicle determined responsive to image-space locations of down-range-separated light spots.

Path sensing using structured lighting

Systems and methods are provided for imaging a region of interest. A structured light source is configured to provide a substantially two-dimensional light curtain within a field of view of the camera formed by at least one projected laser beam at a wavelength detectable by the camera. An image analysis component is configured to detect objects within the region of interest from at least a position of the at least one projected laser beam within the image.

Augmented object detection using structured light

At least one fan beam of light () projected along a path () of a vehicle () generates a corresponding at least one light stripe () along the surface (') of the path (), which is imaged by an imaging system () to generate a pair of stereo image components (&#x2033;) which is processed so as to provide for sensing a physical feature of or along the path (&#x2032;).

Vehicular path sensing system and method

An image of a visual scene, comprising a plurality of pixels. is acquired and an associated range map is either determined therefrom or separately acquired. Elements of the range map comprise distances from the camera for each pixel of the image. In one aspect, either the image or the range map is processed with a connected-components sieve filter that locates clusters of pixels or elements that are connected to one another along either adjacent rows, columns or diagonally. In another aspect, a cross-range value of range-map element is compared with a down-range-responsive cross-range threshold of a boundary of a collision-possible space and the pixel or element is nulled or ignored if associated with a location that is not in the collision-possible space. The collision-possible space is responsive to an operating condition of a vehicle from which the image is acquired.

Method of processing an image of a visual scene

A driver alertness detection system includes an imaging unit configured to image an area in a vehicle compartment where a driver's head is located; an image processing unit configured to receive the image from the imaging unit, and to determine positions of the driver's head and eyes; and a warning unit configured to determine, based on the determined position of the driver's head and eyes as output by the image processing unit, whether the driver is in an alert state or a non-alert state, and to output a warning to the driver when the driver is determined to be in the non-alert state.

System and method for determining driver alertness

Objects in a range map image are clustered into regions of interest responsive to range as determined from either a separate ranging system or from a top-down transformation of a range map image from a stereo vision system. A relatively-central location of each region of interest is transformed to mono-image geometry, and the corresponding portion of an associated mono-image is searched radially outwards from the relatively-central location along a plurality of radial search paths, along which the associated image pixels are filtered using an Edge-Preserving Smoothing filter in order to find an edge of the associated object along the radial search path. Edge locations for each of the radial search paths are combined in an edge profile vector that provides for discriminating the object.

Range-cued object segmentation system and method

For each of a plurality of homogeneity regions in relative pixel space, a deviation associated therewith with respect to a pixel to be filtered is given by a sum of associated difference values, each difference value given by the absolute value of a difference between a value of a pixel to be filtered and that of a neighboring pixel selected in accordance with the selected homogeneity region. The filtered pixel value is responsive to values of the neighboring pixels for the homogeneity region with minimum deviation. The relative pixel locations of each homogeneity region are symmetric relative to a radially-extending axis extending outwards in a corresponding polar direction therealong from a relatively central pixel location to a boundary of the homogeneity region, including all relative pixels intersected by the radially-extending axis, different homogeneity regions being associated with different polar directions.

Method of filtering an image

A range map of a visual scene generated by a stereo vision and associate image processing system, and is filtered to remove objects beyond a region of interest and for which a collision is not possible, and to remove an associated road surface. Objects clustered in range bins are separated by segmentation. A composite range map is generated using principale components analysis and processed with a connected-components sieve filter. Objects are identified using one or more of a harmonic profile and other features using an object recognition processor using a combination of inclusive, exclusive and harmonic networks to generate a classification metric.

Vulnerable road user protection system

A plurality of features determined from at least a portion of an image containing information about an object are processed with an inclusive neural network, and with a plurality of exclusive neural networks, so as to provide a plurality of inclusive probability values representing probabilities that the portion of the image corresponds to at least one of at least two different classes of objects, and for each exclusive neural network, so as to provide first and second exclusive probability values representing probabilities that the portion of the image respectively corresponds. or not. to at least one class of objects. The plurality of inclusive probability values, and the first and second exclusive probability values from each of the exclusive neural networks, provide for identifying whether the portion of the image corresponds, or not, to any of the at least two different classes of objects.

Method of identifying an object in a visual scene

A camera () captures successive images of light stripes () projected onto an object by a light curtain () positioned by a light source positioner (). A background image is subtracted () therefrom, and the resulting image is boosted by binning (), binarized with a thresholding algorithm (), skeletonized (), interpolated () and stored (). Interpolated images are acquired for a plurality of light stripes (). A processor () generates () a 3-D surface model from Cartesian coordinates computed for non-zero camera pixels. A volumetric representation is determined () from the offset of the object surface model relative to a model of a proximate surface, e.g. a seating surface (). The object is classified (), e.g. by a trainable pattern recognitions system, responsive to 3-D shape descriptors () of the 3-D surface model and to the volumetric representation () or portions () thereof. The detection of an occupant () in a vehicle () may be used to control a safety restraint system ().

Occupant detection system

An eye-safe light curtain is generated by a plurality of LED's illuminating a cylindrical lens having a corrugated cylindrical surface. A camera generates a first image of an avoidance zone boundary of an air bag inflator with the light curtain off, and a second image of with the light curtain on along the avoidance zone boundary. The first image is subtracted from the second image, and the resulting difference image is binned to enhance signal strength. A penetration of the avoidance zone is classified from features of the leading edge of the binned image, and the air bag inflator is disabled if the avoidance zone is penetrated by an occupant at risk of injury from the deployment thereof. An image of the light curtain may be recovered for high ambient lighting conditions by adapting the exposure time of the camera responsive to the percentage of pixels in a mask region of the first image having values exceeding a threshold.

Growing community of inventors

South Lyon, MI, United States of America

Gregory Gerhard Schamp