A method and a device for 3D shape matching based on a local reference frame are proposed. After acquiring a 3D point cloud and feature points in the method, the feature point set is projected to a plane, and feature transformation is performed on the projected points by using at least one factor from the distances between the 3D points and the feature points, the distances between the 3D points and the projected points, and the average distances between the 3D points and its 1-ring neighboring points to acquire a point distribution with a larger variance in a certain direction than the projected point set, and the local reference frame is determined based on the transformed point distribution. The 3D local feature descriptor established based on this local reference frame can encode the 3D local surface information more robustly, so as to obtain a better 3D shape matching effect.

A system is configured for converting an unstandardized architecture diagram into a braille language diagram is disclosed. The system receives the unstandardized architecture diagram which includes a plurality of architecture components. The system receives a standardized model that includes features to depict the architecture components of the unstandardized architecture diagram in a standard format. The system determines the architecture components, their connections, and their sequences from the unstandardized architecture diagram. The system determines the features to depict the architecture components of the unstandardized architecture diagram in the standard format. The system applies the identified features on the identified architecture components in the unstandardized architecture diagram. The system determines a standardized graphical representation of the unstandardized architecture diagram. The system converts the standardized graphical representation of the unstandardized architecture diagram into a braille language diagram.

An ECU captures an image of an imaging region around an own vehicle, and acquires image data, the imaging region being configured by a plurality of imaging areas. The ECU determines whether the object is present in the imaging areas, on the basis of detection results of an object present around the own vehicle. The ECU selects a target area to be displayed in an easy-to-see state from among the plurality of imaging areas, on the basis of determination results. The ECU reduces and corrects the image data of each imaging area such that an image of the target area is displayed in the easy-to-see state compared to an image of each imaging area to generate display image data.

According to one embodiment, an overhead image generation apparatus includes: a plurality of cameras mounted to a vehicle; an image processor that takes in images of respective cameras, generates, for respective cameras, overhead images that have been subjected to viewpoint conversion processing based on calibration data of the cameras and virtual viewpoint/line-of-sight information, and generates a synthesized overhead view by connecting the overhead images at their boundaries; and a display device that displays the synthesized overhead view generated by the image processor, wherein a proportion of a shape of the overhead image in a height direction is changed in proportion to an arrangement height of each camera.

A computer completes a data image analysis task. The computer receives a machine learning (ML) model trained for use with image data content characterized by first context. The computer receives an evaluation image dataset having evaluation image data content characterized by a second context. The computer receives a request to complete an image data analysis task for the evaluation image dataset using the ML model. The computer compares the contexts to determine whether the contexts are similar and whether the evaluation image dataset is compatible with the ML model. If the evaluation dataset is incompatible with the ML model, the computer uses the generative model to generate a ML model compatible synthetic image dataset based on the evaluation dataset. The computer applies the ML model to the synthetic image dataset to provide an answer for the data image analysis task; the computer delivers the answer to a user interface.

A method for multi-camera self-supervised depth evaluation is described. The method includes training a self-supervised depth estimation model and an ego-motion estimation model according to a multi-camera photometric loss associated with a multi-camera rig of an ego vehicle. The method also includes generating a single-scale correction factor according to a depth map of each camera of the multi-camera rig during a time-step. The method further includes predicting a 360° point cloud of a scene surrounding the ego vehicle according to the self-supervised depth estimation model and the ego-motion estimation model. The method also includes scaling the 360° point cloud according to the single-scale correction factor to form an aligned 360° point cloud.

An image generation device performs coordinate transformation, based on a coordinate transformation table, to a two-dimensional first image having a span in a horizontal direction and a vertical direction and acquired by overlooking and imaging an object from a first viewpoint at a first depression angle, and generates and outputs a second image which was obtained by overlooking the object from a second viewpoint which is different from the first viewpoint at a second depression angle which is different from the first depression angle. The coordinate transformation table is a table for transforming coordinates of a plurality of first selected pixels selected from a plurality of first pixels constituting the first image into coordinates of second selected pixels corresponding to a plurality of second pixels constituting the second image.

Examples disclosed herein obtain first image data and the second image data for a foveated image frame to be displayed on a display, the first image data to have a first resolution and the second image data to have a second resolution lower than the first resolution. Disclosed examples also up-sample the second image data based on first metadata to generate up-sampled second image data, the up-sampled second image data to have the first resolution, and combine the first image data and the up-sampled second image data based on second metadata. Disclosed examples further perform, based on third metadata, a combination of at least two different filter operations on an overlap region including a portion of the first image data and a portion of the up-sampled second image data to generate the foveated image frame, the third metadata to specify a width in pixels of the overlap region.

An image processing apparatus includes a restrictive condition storage unit in which at least one restrictive condition, which is to be applied to an image to be output and acquired from a subject, is stored, an accepting unit that accepts an image that is obtained by shooting the subject and has at least one field, an image changing unit that applies the at least one restrictive condition to the at least one field of the image accepted by the accepting unit, changes the at least one field so that it satisfies the at least one restrictive condition, and acquires at least one new field, and an image output unit that outputs the at least one field acquired by the image changing unit, enabling an image having an overall balance to be output.

According to an image transmission method, a pre-processed image is inputted into a compression system, which automatically localizes at least one primary feature in the pre-processed image and then automatically determines a position within the pre-processed image of at least one landmark corresponding to the at least one primary feature. It then automatically foveates the pre-processed image to emphasize portions of the pre-processed image within a pre-determined measure of closeness to the landmark(s) and transmits the foveated image to at least one recipient system, which reconstructs a received version of the pre-processed image by inverse foveation. In one embodiment, an incompressible fluid mechanics model is used to determine the degree of foveation at different points in the pre-processed image.