There is provided an image processing apparatus. A line segment detecting unit detects line segments from an image. An obtainment unit obtains distance information indicating a distance in a depth direction of the image. A selecting unit selects a line segment for vanishing point detection from the line segments detected by the line segment detecting unit. A vanishing point detecting unit detects a vanishing point of the image on a basis of the line segment for vanishing point detection. For each line segment detected by the line segment detecting unit, the selecting unit selects the line segment in a case where the distance indicated by the distance information is in a monotonically changing trend along the line segment.

A robot includes an image sensor that captures the environment of a storage site. The robot visually recognizes regularly shaped structures to navigate through the storage site using various object detection and image segmentation techniques. In response to receiving a target location in the storage site, the robot moves to the target location along a path. The robot receives the images as the robot moves along the path. The robot analyzes the images captured by the image sensor to determine the current location of the robot in the path by tracking a number of regularly shaped structures in the storage site passed by the robot. The regularly shaped structures may be racks, horizontal bars of the racks, and vertical bars of the racks. The robot can identify the target location by counting the number of rows and columns that the robot has passed.

An image recognition system and method for a vehicle, including at least two camera units, each being configured to record an image of a road in the vicinity of the vehicle and to provide image data representing the respective image of the road, a first image processor configured to combine the image data provided by the at least two camera units into a first top-view image. The first top-view image is aligned to a road image plane, a first feature extractor configured to extract lines from the first top-view image, a second feature extractor configured to extract an optical flow from the first top-view image and a second top-view image, generated before the first top-view image by the first image processor, and a curb detector configured to detect curbs in the road based on the extracted lines and the extracted optical flow and provide curb data representing the detected curbs.

Certain aspects relate to systems and techniques for efficiently recording captured plenoptic image data and for rendering images from the captured plenoptic data. The plenoptic image data can be captured by a plenoptic or other light field camera. In some implementations, four dimensional radiance data can be transformed into three dimensional data by performing a Radon transform to define the image by planes instead of rays. A resulting Radon image can represent the summed values of energy over each plane. The original three-dimensional luminous density of the scene can be recovered, for example, by performing an inverse Radon transform. Images from different views and/or having different focus can be rendered from the luminous density.

A system segments a set of images of a property to identify a type of damage to the property. The system receives, from an image capturing device, a digital image of a roof or other feature of the property. The system processes the image to identify a set of segments, in which each segment corresponds to a piece of the feature, such as a tab or tooth of a shingle on the roof. The system saves a result of the processing to a data file as a segmented image of the property, and it uses the segmented image to identify a type of damage to the property.

In a method for making a measurement of the hip in an ultrasound image, an ultrasound image of the hip is obtained. A spatial coherence map associated with one or more lags of the ultrasound waves associated with the ultrasound image is also obtained. Then a quality metric is determined based on the ultrasound image of the hip and the spatial coherence map. The quality metric indicates the suitability of the ultrasound image for making the measurement of the hip. If the quality metric indicates that the ultrasound image is above a threshold quality, then the method then comprises indicating that the ultrasound image is suitable for making the measurement of the hip.

A method for creating a plane panorama from point cloud data using Hough transformations is disclosed. The method involves converting the three-dimensional point cloud into a two-dimensional histogram with bins grouping neighboring points, and performing a Hough transformation on the histogram. The resulting transformed data is segmented and the method searches the segments iteratively for a major line, followed by lines that are orthogonal, diagonal, or parallel to the major line, and discards outlying data in each bin as lines are identified. The detected lines are connected to form planes, and the planes are assembled into a hole- and gap-filled panorama. The method may also use an algorithm such as a Random Sample Consensus (RANSAC) algorithm to detect a ground plane.

A method and a device for image stabilization of an image sequence of one or more cameras of a vehicle are disclosed. Images of the surroundings of the vehicle are recorded and image points of the recorded images are projected to a unit sphere, where the projection includes applying a lens distortion correction. For each of the one or more cameras, a vanishing point is calculated using the projected image points on the unit sphere and a motion of the vanishing point on the unit sphere is tracked. The motion of the vanishing point is used to calculate a corrected projection to a ground plane.

A method for detecting defects on a sample based on a defect inspection apparatus is provided. In the method, an image data set that includes defect data and non-defect data is organized. A convolutional neural network (CNN) model is defined. The CNN model is trained based on the image data set. The defects on the sample are detected based on inspection data of the defect inspection apparatus and the CNN model. The sample includes uniformly repeating structures, and the inspection data of the defect inspection apparatus is generated by filtering out signals of the uniformly repeating structures of the sample.

A lane mark recognition device detects an edge located in a proximal area in front of the vehicle, and determines a lane mark candidate on the basis of the detected edge and an edge that is located in a distal area farther than the proximal area in front of the vehicle and that is continuous with the edge. Therefore, an edge of another vehicle (leading vehicle) or the like located in the distal area in a captured image is not detected as a lane mark candidate (is excluded as a non lane mark candidate).