Abstract:
Disclosed herein is an image deep learning model training method. The method includes sampling a twin negative comprising a first negative sample and a second negative sample by selecting the first negative sample with a highest similarity out of an anchor sample and a positive sample constituting a matching pair in each class and by selecting the second negative sample with a highest similarity to the first negative sample, and training the samples to minimize a loss of a loss function in each class by utilizing the anchor sample, the positive sample, the first and second negative samples for each class. The first negative sample is selected in a different class from a class comprising the matching pair, and the second negative sample is selected in a different class from classes comprising the matching pair and the first negative sample.
Abstract:
An apparatus and a method for displaying a panoramic image using a look-up table (LUT) are disclosed, including generating an LUT may include determining first geometric correction information to transform an input domain pixel coordinate system of an input image to a panorama domain pixel coordinate system of a panoramic image, determining second geometric correction information to transform an output domain pixel coordinate system of an output image of the panoramic image to the panorama domain pixel coordinate system of the panoramic image, determining third geometric correction information to transform the output domain pixel coordinate system of the output image to the input domain pixel coordinate system of the input image based on the first geometric correction information and second geometric correction information, and generating an LUT that maps the output domain pixel coordinate system of the output image to the input domain pixel coordinate system of the input image.
Abstract:
A method of receiving content in a client is provided. The method may include receiving, from a server, a spatial set identifier (ID) corresponding to a tile group including at least one tile, sending, to the server, a request for first content corresponding to metadata, and receiving, from the server, the first content corresponding to the request.
Abstract:
An apparatus and method for generating and consuming a three-dimensional (3D) data format to generate a realistic panoramic image are provided. The apparatus may include an image preprocessing unit to search for a matching point between images captured by a plurality of cameras, and to extract, as image information, at least one of a depth value, a texture value and object division information from each of the captured images, an image information structuring unit to structure 3D data to use the extracted image information to generate a realistic image, a 3D data format storage unit to store format information of the structured 3D data in a database (DB), realistic image generating unit to generate a realistic panoramic image using the stored format information of the 3D data, and a realistic image rendering unit to perform rendering on the generated realistic panoramic image.
Abstract:
Rate adaptation is carried out using bit error rate (BER) to enable effective multimedia transmission. The BER can be estimated using signal strength in a MAC layer and modulation information (FIGS. 7-9), and can be compatibly used in different wireless networks by means of message standardization.
Abstract:
Disclosed are a method and an apparatus for generating a panoramic image based on an image quality. The method of generating a panoramic image includes extracting a matching point to connect a base image captured by a first camera and a reference image captured by a second camera, geometrically transforming the reference image by determining a homography between the base image and the reference image, determining a change in image quality of the geometrically transformed reference image, and generating a panoramic image in which the geometrically transformed reference image is connected to the base image based on the determined change in image quality.
Abstract:
Provided is an apparatus and method for detecting a key point using a high-order Laplacian of Gaussian (LoG) kernel. The high-order LoG kernel is generated based on an LoG operator which is calculated by sequentially differentiating an LoG operator with respect to x and y of an image. A scale space is generated based on the high-order LoG kernel and the key point is detected by comparing a current pixel in the scale space to pixels adjacent to the current pixel.
Abstract:
Provided is a geometric correction apparatus and method based on a recursive Bezier patch sub-division. A geometric correction method may include: receiving, from a camera, a first image that is obtained by photographing a black screen that is projected by a projector onto a projection surface; receiving, from the camera, a second image that is obtained by photographing a predetermined pattern that is projected by the projector onto the projection screen; generating a third image by subtracting the first image from the second image; and performing geometric correction with respect to the predetermined pattern to correct a distortion between the predetermined pattern and the third image.