Abstract:
A Real- Time, Semi-Automatic Method for Discriminant Track Initialization in Thermal Imagery Track initialization is very critical for tracking since it determines what to track for the tracker. Therefore, any insignificant or false information, i.e. parts of objects similar to common background or patches from background, may result in redundant features or deceptive appearance that can cause premature track losses. To achieve long-term tracking, a real-time, semi-automatic track initialization methodology for IR images is proposed which takes a single image coordinate as input, then generates target bounding box for the most salient segment. The present invention is designed for real-time applications in which erroneous user input is unavoidable. Therefore, the present invention is user friendly due to its error compensation capability, i.e. exactly same track initialization can be achieved even if the given input coordinate scatters; and also boosts performance of several tracking algorithms in terms of the number of frames in which successful tracking is achieved since better discriminative targets are obtained. Very low computational cost and requirement of only a point coordinate as input in the vicinity of the target make this approach preferable in real-time tracking applications.
Abstract:
A method in a radio communications network is described of using image data for positioning a first wireless device. The method comprises obtaining image data comprising at least two image objects and an angle between lines formed between each of the objects and a reference point. The method further comprises extracting the at least two image objects from the obtained image data, determining, from the obtained image data, the angle between the lines formed between each of the objects and the reference point, and using the at least two image objects and/or the image data for positioning the first wireless device based on the determined angle. The method is performed by at least one of: a network node, the first wireless device, a camera device, a second wireless device, an image processing node and a server, each operating in the radio communications network.
Abstract:
The present disclosure provides notably a calibration system suitable for in-flight calibration of a sensor payload. The calibration system comprises an emitting object being configured for emitting in a first emitting spectral band and in a second emitting spectral band a predetermined pattern comprising a plurality of lighted areas on a homogeneous background; and a collimation optical unit configured for setting the emitting object at infinity.
Abstract:
Method and processor for searching for a target within video data comprising the steps of receiving a target selected from within video data. Identifying a current selection of target matches for the selected target within further video data. Ranking the current selection of target matches. Receiving a signal confirming or rejecting one or more of the ranked target matches. Identifying a further selection of target matches for the confirmed target matches from the further video data. Indicating portions of the further video data containing the further selection of target matches.
Abstract:
Search terms are derived automatically from images captured by a camera equipped cell phone, PDA, or other image capturing device, submitted to a search engine to obtain information of interest, and at least a portion of the resulting information is transmitted back locally to, or nearby, the device that captured the image.
Abstract:
The present invention is related to a method of identifying at least one sequence of target regions on a plurality of macromolecules to test, each target region being associated with a tag and said macromolecules having underwent linearization according to a predetermined direction, wherein said method comprises performing by a processor (11) of equipment (10) the following steps: (a) receiving from a scanner (2) being sensitive to said tags, at least one sample image depicting said macromolecules as curvilinear objects sensibly extending according to said predetermined direction; (b) Generating a binary image from the sample image; (c) For at least one template image, and for each sub-area of the binary image having the same size as the template image, calculating a correlation score between the sub-area and the template image; (d) For each sub-area of the binary image for which the correlation score with a template image is above a first given threshold, selecting the corresponding sub-area of the sample image; (e) For at least one reference code pattern, and for each selected sub-area of the sample image, calculating an alignment score between the sub- area and the reference code pattern, said reference code pattern being defined by a given sequence of tags; (f) For each selected sub-area of the sample image for which the alignment score with a reference code pattern is above a second given threshold, identifying each target region depicted in said selected sub- area among the target regions associated with the tags defining said reference code pattern; (g) Outputting the different sequence(s) of identified target regions.
Abstract:
Disclosed is a method for object tracking, comprising: determining a region of interest (ROi) in a first frame of a video sequences, wherein the ROi is centered at a ground truth target location for objects to be tracked; feeding the determined R0I forward through a first CNN (convolutional network) to obtain a plurality of first feature maps in a higher layer of the CNN and a plurality of second feature maps in a lower layer of the first CNN, wherein the first CNN is pre-trained on an image classification task such that the first feature maps include more semantic features to determine a category for objects to be tracked in the video sequences, while the second feature maps carry more discriminative information to separate the objects from distracters with similar appearance; selecting a plurality of feature maps from the first and second feature maps, respectively by training a second CNN and a third CNN (sel-CNNs) with the first feature maps and the second feature maps, respectively; predicting, based on the selected first and second feature maps, two target heat maps indicating a target location for said objects in the current frame, respectively; and estimating, based on the two predicated target heat maps, a final target location for the object in the current frame.