An image processing device 100 of the present invention includes an image generation means 121 for generating a difference image representing a difference between an object image that is an image including an area from which a mobile body is to be detected and a corresponding image that is another image including an area corresponding to the area of the object image, and a detection means 122 for detecting the mobile body from the object image on the basis of the object image, the corresponding image, and the difference image.

A method for ascertaining a starting position of a vehicle for a localization of the vehicle using a control unit. In the method, measurement data are received from an odometry sensor system and/or a GNSS sensor system of the vehicle, a first position and an uncertainty range of the first position are determined based on the measurement data received, at least one map section of a feature map containing a plurality of stored features is received, the map section having a position and extent which is superimposed on the first position and the uncertainty range, measurement data are received from a LiDAR sensor system, a radar sensor system and/or a camera sensor system and static features are extracted from the measurement data received, a first starting position of the vehicle is ascertained by comparing the static features extracted from measurement data with features stored in the map section.

Automated methods and systems are disclosed, including a method comprising: obtaining a first three-dimensional-data point cloud of a horizontal surface of an object of interest, the first three-dimensional-data point cloud having a first resolution and having a three-dimensional location associated with each point in the first three-dimensional-data point cloud; capturing one or more aerial image, at one or more oblique angle, depicting at least a vertical surface of the object of interest; analyzing the one or more aerial image with a computer system to determine three-dimensional locations of additional points on the object of interest; and updating the first three-dimensional-data point cloud with the three-dimensional locations of the additional points on the object of interest to create a second three-dimensional-data point cloud having a second resolution greater than the first resolution of the first three-dimensional-data point cloud.

A method of determining kinetic information may include: receiving a plurality of raw information related to a plurality of objects using a radar device provided in a vehicle; obtaining, by analyzing the plurality of raw information, a plurality of candidate kinetic information related to the vehicle; estimating, through spatial filtering, current first kinetic information related to the vehicle from the plurality of candidate kinetic information; and correcting, using a kinetic model, the estimated current first kinetic information based on current first kinetic information, wherein the current first kinetic information is predicted from previous first kinetic information related to the vehicle using a kinetic model.

A system and method for identifying damage to an embankment includes acquiring satellite imagery of an area of the embankment, generating a set of input data from the satellite imagery, removing at least one anomaly in the set of input data to obtain a cleaned set of input data, and identifying the damage by determining a dam motion area indicative of ground motion in the embankment from the cleaned set of input data and determining an anomalous vegetation area and an anomalous wetness area indicative of seepage in the embankment from the cleaned set of input data.

This document describes Kurtosis based pruning for sensor-fusion systems. Kurtosis based pruning minimizes a total quantity of comparisons performed when fusing together large sets of data. Multiple candidate radar tracks may possibly align with one of multiple candidate visual tracks. For each candidate vision track, a weight or other evidence of matching is assigned to each candidate radar track. An inverse of matching errors between each candidate vision and each candidate radar track contributes to this evidence, which may be normalized to produce, for each candidate vision track, a distribution associated with all candidate radar tracks. A Kurtosis or shape of this distribution is calculated. Based on the Kurtosis values, some candidate radar tracks are selected for matching and other remaining candidate radar tracks are pruned. The Kurtosis aids in determining how many candidates to retain and how many to prune. In this way, Kurtosis based pruning can prevent combinatorial explosions due to large-scale matching.

Examples disclosed herein relate to an autonomous driving system in an vehicle. The autonomous driving system includes a radar system configured to detect a target in a path and a surrounding environment of the vehicle and produce radar data with a first resolution that is gathered over a continuous field of view on the detected target. The system includes a super-resolution network configured to receive the radar data with the first resolution and produce radar data with a second resolution different from the first resolution using first neural networks. The system also includes a target identification module configured to receive the radar data with the second resolution and to identify the detected target from the radar data with the second resolution using second neural networks. Other examples disclosed herein include a method of operating the radar system in the autonomous driving system of the vehicle.

A contextual visual-based synthetic-aperture radar (SAR) target detection method and apparatus, and a storage medium, belonging to the field of target detection is described. The method includes: obtaining an SAR image; and inputting the SAR image into a target detection model, and positioning and recognizing a target in the SAR image by using the target detection model, to obtain a detection result. In the present disclosure, a two-way multi-scale connection operation is enhanced through top-down and bottom-up attention, to guide learning of dynamic attention matrices and enhance feature interaction under different resolutions. The model can extract the multi-scale target feature information with higher accuracy, for bounding box regression and classification, to suppress interfering background information, thereby enhancing the visual expressiveness. After the attention enhancement module is added, the detection performance can be greatly improved with almost no increase in the parameter amount and calculation amount of the whole neck.

The embodiments of the present disclosure provide a method for a flow velocity detection. The method may include obtaining image data; determining, based on the image data, a parameter of at least one detection point, the parameter being related to a phase change; and determining a first flow velocity of the at least one detection point based on the parameter related to the phase change and a location relationship among the at least one detection point, at least one transmission point, and a plurality of receiving points.

A method may include receiving a group of images taken by a camera over time in an environment, in which the camera may be oriented within the environment to capture images of an object in a substantially same direction as a launch direction of the object, and the group of images including a first image and a second image. The method may further include: identifying a first position of the object in the first image; identifying a second position of the object in the second image; generating a flight vector based on the first position of the object and the second position of the object; and determining one or more flight parameters using the flight vector. Additionally, the method may include: generating a simulated trajectory of the object based on the flight parameters; and providing the simulated trajectory of the object for presentation in a graphical user interface.