Abstract: An image processing device includes an image sensor including a unit block including a plurality of pixels arranged adjacent to each other, wherein a plurality of nano-posts are arranged on the unit block, and a processor that processes an input image acquired through the image sensor.

Abstract: A processor-implemented method including masking a point cloud according to a mask, the mask being formed as a grid, training a first auto-encoder to reconstruct coordinates and densities of masking points of the point cloud, the masking points respectively corresponding to masking grid cells of the mask, from residual points of the point cloud, the residual points respectively corresponding to residual grid cells of the mask, and training a second auto-encoder to reconstruct colors, the coordinates, and the densities of the masking points from an image corresponding to the point cloud.

Abstract: An image processing device includes an image sensor including a unit block including a plurality of pixels arranged adjacent to each other, wherein a plurality of nano-posts are arranged on the unit block, and a processor that processes an input image acquired through the image sensor.

Abstract: A scene flow estimation method and apparatus are provided. The scene flow estimation method includes receiving a first feature pyramid and a second feature pyramid by encoding a first frame and a second frame of an input image through the same encoder, extracting a depth feature based on the first feature pyramid, extracting a motion feature based on the first feature pyramid and the second feature pyramid, generating an overall feature based on the depth feature and the motion feature, and estimating a scene flow based on the overall feature.

Abstract: A scene flow estimation method and apparatus are provided. The scene flow estimation method includes receiving a first feature pyramid and a second feature pyramid by encoding a first frame and a second frame of an input image through the same encoder, extracting a depth feature based on the first feature pyramid, extracting a motion feature based on the first feature pyramid and the second feature pyramid, generating an overall feature based on the depth feature and the motion feature, and estimating a scene flow based on the overall feature.

Abstract: Disclosed is a method for intelligently displaying a sports game video for a multimedia mobile terminal, the method including the steps of: determining if a camera shot existing in the sports game video is a long-shot; when the camera shot is determined as a long-shot, determining a ROI within an image frame of the sports game video; and when the camera shot is determined as a non-long-shot, displaying the image frame on the mobile terminal, and, when the camera shot is determined as a long-shot, enlarging and displaying the ROI on the mobile terminal.

Abstract: A method, medium, and system generating a depth map of a video image are provided. The depth map generating method extracts the ground of a video image other than an object from the video image, classifies the video image as a long shot image or a non-long shot image based on a distribution value of the extracted ground, calculates a depth value gradually varied along a predetermined direction of the extracted ground when the video image corresponds to the long shot image and calculates a depth value based on the object when the video image corresponds to the non-long shot image. Accordingly, a sense of space and perspective can be effectively given to even a long shot image in which the ground occupies a large part of the image and a stereoscopic image recognizable by a viewer can be generated even if rapid object change is made between scenes in a video image.

Abstract: Disclosed is a method for intelligently displaying a sports game video for a multimedia mobile terminal, the method including the steps of: determining if a camera shot existing in the sports game video is a long-shot; when the camera shot is determined as a long-shot, determining a ROI within an image frame of the sports game video; and when the camera shot is determined as a non-long-shot, displaying the image frame on the mobile terminal, and, when the camera shot is determined as a long-shot, enlarging and displaying the ROI on the mobile terminal.

Abstract: An inter mode for encoding a video macroblock is selected. An initial analysis on the macroblock homogeneity is performed to quickly reduce the number of inter modes for which motion estimation and rate distortion calculation are to be performed. An inter mode is selected based on these calculations and sub-pixel motion estimation, at various sub-pixel interpolation granularities, is performed only on the selected inter mode. Sub-pixel motion estimation may be skipped for macroblocks that exhibit relatively low distortion at the integer pixel level.

Abstract: Disclosed is a method for intelligently displaying a sports game video for a multimedia mobile terminal, the method including the steps of: determining if a camera shot existing in the sports game video is a long-shot; when the camera shot is determined as a long-shot, determining a ROI within an image frame of the sports game video; and when the camera shot is determined as a non-long-shot, displaying the image frame on the mobile terminal, and, when the camera shot is determined as a long-shot, enlarging and displaying the ROI on the mobile terminal.

Abstract: Transform based distortion cost estimation, which may be used to calculate a rate distortion cost associated with a mode for encoding a macroblock, is described. In one embodiment of the invention, a distortion value for a particular mode is estimated within the transform domain, which allows for the elimination of both inverse transformation and inverse motion functions in this calculation. A spatial domain residual of an encoding mode is estimated by identifying a difference, within the transform domain, between a motion compensated prediction residual of a macroblock and its corresponding reconstructed signal. The estimated spatial domain residual may then be used in distortion matrix computations to estimate a distortion level, within the transform domain, for an encoding mode.

Abstract: An inter-mode for encoding a video macroblock is selected based on a comparison of at least two rate-distortion values associated with inter-modes. The number of potential inter-modes for encoding the video macroblock is initially reduced based on the rate-distortion calculation of the SKIP mode between the current macroblock and the collocated macroblock (corresponding to a 0,0 motion vector). Motion estimation is performed on the remaining inter-modes to identify reference blocks and motion vectors. The number of calculations performed in identifying the reference blocks and motion vectors is potentially reduced because associations between inter-modes are recognized and leveraged.

Abstract: A prediction mode for encoding data is selected using a reduced number of rate-distortion cost computations. This reduction in rate-distortion cost computations is caused by filtering the number of potential intra prediction modes based on two criteria. First, the number of potential prediction modes is reduced based on at least one characteristic of a quantization procedure performed during the encoding procedure. Second, the number of potential prediction modes is reduced based on an error value calculated for each of the potential prediction modes.

Abstract: A method for smoothing artificial discontinuities between image blocks associated with digital data comprises reconstructing a block-based pixel representation of image blocks associated with digital data. If the difference between adjacent image blocks of the block-based pixel representation is less than or equal to a quantization parameter, then the method includes modifying boundary pixel values to define at least one additional frame, and displaying the at least one additional frame and an original frame in an alternating fashion so as to visually achieve smooth block boundaries between the adjacent image blocks. The method may be performed by a suitable device and/or may be embodied in an instruction set executable by such device.

Abstract: A method, medium, and system generating a depth map of a video image are provided. The depth map generating method extracts the ground of a video image other than an object from the video image, classifies the video image as a long shot image or a non-long shot image based on a distribution value of the extracted ground, calculates a depth value gradually varied along a predetermined direction of the extracted ground when the video image corresponds to the long shot image and calculates a depth value based on the object when the video image corresponds to the non-long shot image. Accordingly, a sense of space and perspective can be effectively given to even a long shot image in which the ground occupies a large part of the image and a stereoscopic image recognizable by a viewer can be generated even if rapid object change is made between scenes in a video image.

Abstract: A method for a matching technique for detecting copies of a video clip is provided. The method includes comparing temporal trails of intensity distributions between frames of the video clip and frames of a target video. A computer readable media, a computer system and integrated circuit for video copy detection are also provided.

Abstract: A two-step matching technique is embodied in a video-copy-detection algorithm that detects copies of video sequences. The two-step matching technique uses ordinal signatures of frame partitions and their differences from partition mean values. The algorithm of this invention is not only robust to intensity/color variations it can also effectively handle various format conversions, thereby providing robustness regardless of the video dynamics of the frame shots.

Abstract: A 3-D dot code is embedded in a printed medium using an embedder that embeds data directly into the halftone dots of a selected file that is ready for printing. An extractor extracts the embedded bits by scanning the printed medium and then interprets the embedded data. A bleed-through elimination module in the extractor is able to eliminate adverse affects on the scanned input of the embedded data caused by bleed-through from the print on the opposite side of the page. The extractor may also include a module that compensates for a misaligned scanned input by automatically detecting the rotation and translation, and then registering the image so that the upper-left corner is located in a certain position. This operation expedites subsequent data retrieval, in which data is read by block matching.

Abstract: A method for extracting an object of interest from an image is provided. The method initiates with defining an image feature space based upon frequency information. Then, the image feature space is filtered to smooth both focused regions and defocused regions while maintaining respective boundaries associated with the focused regions and the defocused regions. The filtered image feature space is manipulated by region merging and adaptive thresholding to extract an object-of-interest. A computer readable media, an image capture device and an image searching system are also provided.

Abstract: Disclosed is a method for intelligently displaying a sports game video for a multimedia mobile terminal, the method including the steps of: determining if a camera shot existing in the sports game video is a long-shot; when the camera shot is determined as a long-shot, determining a ROI within an image frame of the sports game video; and when the camera shot is determined as a non-long-shot, displaying the image frame on the mobile terminal, and, when the camera shot is determined as a long-shot, enlarging and displaying the ROI on the mobile terminal.