Telefonaktiebolaget Lm Ericsson (publ)

Manish Sonal

Volodya Grancharov

Alfredo Fanghella

Sigurdur Sverrisson

José Araújo

Steven Combes

A method performed by an image registration entity includes obtaining a matching between a first set of objects in a first image of a scene and a second set of objects in a second image of the scene, and obtaining a first set of key-points from the first image and a second set of key-points from the second image. Image registration is performed by matching the first set of key-points to the second set of key-points. Those of the first set of key-points that are mapped to objects in the first set of objects and that have matching objects in the second set of objects are restricted to only be matched to those of the second set of key-points that are mapped to any of the matching objects in the second set of objects. The matching between key-points resulting from the image registration is applied when constructing an image representation of the scene.

Image registration based on key-point matching

A method () for producing a reduced point cloud comprising an object of interest (OOI) from an original point cloud comprising the OOI. The method includes obtaining (s) the original point cloud; obtaining (s) a set of N images, each of the N images comprising an image of the OOI from a unique position relative to the OOI; for each one of the N images, defining (s) an area of interest in the image that obtaining a set of N images, each of the N images that includes the OOI, thereby defining N areas of interest; for each point included in the original point cloud, determining (s), for each one of the N areas of interest, whether the point is located in the area of interest; for each point included in the original point cloud, determining (s) a first metric for the point based on the total number of the N areas of interest in which the point is determined to be located; and for each point included in the set of points, determining (s) whether or not to include the point in the reduced point cloud based on the first metric for the point.

Three-dimensional modeling

There is provided mechanisms for visual object detection in a sequence of images. A method is performed by a visual object detector (). The method comprises obtaining (S) a sequence of images of a scene. The sequence of images at least comprises a current image of the scene and a previous image of the scene. The method comprises extracting (S) a set of objects from the sequence of images by performing visual object detection in the sequence of images. Performing the visual object detection in at least part of the current image is conditioned on a set of conditions being fulfilled. The set of conditions at least pertains to an image-wise descriptor classification score computed for at least one of the previous image and the current image and pertaining to which type of content the scene comprises, and an image overlapping score pertaining to how much overlap in image area there is between the previous image and the current image. The method comprises constructing (S) an image representation of the scene using the extracted set of objects.

Visual object detection in a sequence of images

There is provided mechanisms for joint visual object detection and object mapping to a 3D model. A method is performed by an image processing device. The method includes obtaining a first sequence of digital images of a scene as captured by a first image capturing unit, and obtaining a second sequence of digital images of the scene as captured by a second image capturing unit. The second sequence of digital images is time-wise synchronized with the first sequence of digital images by being captured time-wise in parallel with the first sequence of digital images. The first image capturing unit has a narrower field of view than the field of view of the second image capturing unit. The first image capturing unit and the second image capturing unit have a known spatial relation. The method includes performing joint visual object detection and object mapping to the 3D model.

Joint visual object detection and object mapping to a 3D model

Object parts () are detected in a picture using object detector(s) () and part location representations () are generated for the detected object parts (). The size of an object () comprising object parts () is estimated based on a geometric model and the part location representations (). Search locations () in the picture for a search window () having a size based on the estimated size are determined based on the part location representations (). The search locations () are then processed by identifying any detected object part () that is within the search window () positioned at the search location (). A homography is estimated by minimizing an error between mapped object part(s) from the geometric model and the identified detected object part(s) (). If the error is smaller than a threshold value, an object location representation is determined for the object ().

Object location determination

There is provided mechanisms for identifying decreasing object detectability for a video stream. A method is performed by a controller. The method includes receiving the video stream over a communication channel. The method includes receiving first visual object detection (VOD) data of the video stream. The first VOD data has been obtained from the video stream before is sent over the communication channel towards the controller. The method includes obtaining second VOD data of the video stream. The second VOD data is obtained from the video stream after is received by the controller. The method includes determining, by comparing the first VOD data to the second VOD data, whether any object present in the first VOD data has a decreased object detectability in the second VOD data, thereby identifying the decreasing object detectability for the video stream. The method includes issuing a notification when the decreasing object detectability has been identified.

Identification of decreased object detectability for a video stream

A method for detecting a change in an area of interest. The method includes: 1) forming a pointcloud that comprises: i) a first set of points derived from a first set of one or more images and ii) a second set of points derived from a second set of one or more images; 2) for each point included in the pointcloud, assigning a similarity score to the point; 3) for at least each point from the pointcloud that is included in a particular set of points (e.g., for each point in the pointcloud that is assigned a similarity score that exceeds a threshold (e.g. 0)), determining a value Nt and a value Nnew and adjusting the similarity score assigned to the point based on Nt and Nnew; and 4) using the adjusted similarity scores to determine whether the area of interest has changed.

Growing community of inventors

Sollentuna, Sweden

Manish Sonal