Abstract: A neural network-based image processing method and apparatus are provided. The method includes receiving an input image having a first resolution, and estimating a preview image of the input image having a second resolution that is lower than the first resolution by using a neural network model.

Abstract: A method for training a model using a meta-learning includes training a base learning model; training a loss meta-learning model used as a loss function of the base learning model for each task; and training a meta-learning model used to optimize a parameter of the loss meta-learning model.

Abstract: A processor-implemented method with image processing includes: providing retouch result candidates of an input image to a user in response to applying vector value candidates to a style vector; determining a vector value of the style vector based on a selection of the user for the retouch result candidates; determining an adjustment parameter set corresponding to the determined vector value of the style vector; and generating a retouch result by adjusting the input image based on the adjustment parameter set.

Abstract: A method with image restoration includes: receiving an input image and a first task vector indicating a first image effect among candidate image effects; extracting a common feature shared by the candidate image effects from the input image, based on a task-agnostic architecture of a source neural network; and restoring the common feature to a first restoration image corresponding to the first image effect, based on a task-specific architecture of the source neural network and the first task vector.

Abstract: A processor-implemented object detection method is provided. The method receives an input image, generates a latent variable that indicates a feature distribution of the input image, and detects an object in the input image based on the generated latent variable.

Abstract: A processor-implemented object detection method is provided. The method receives an input image, generates a latent variable that indicates a feature distribution of the input image, and detects an object in the input image based on the generated latent variable.

Abstract: A method of processing frames obtained by multiple exposures may include categorizing, by at least one processor, a plurality of frames captured by multiple exposures into one of a reference frame and a target frame, compensating for global motion resulting from motion of a capturing apparatus capturing an object, and compensating for local motion resulting from motion of the object, based on a joint histogram between the reference frame and the target frame.

Abstract: An apparatus and method creating a ghost-free High Dynamic Range Image (HDRI) based on filtering are provided. It is possible to effectively prevent a ghost phenomenon from occurring when a single HDRI is created from a plurality of LDRIs, by defining a ghost area using a probability based on a global transfer function indicating a relationship for intensities of several frames, rather than searching for or identifying a ghost area in a single or each image.

Abstract: Disclosed are a system and a method extensible for performing, in real-time, stereo snatching for calculating depth images with a result of searching for points of similarity by using images taken with two cameras. The system includes a coordinate creating module, a census transform module, a delay XOR calculation module, a stereo matching module, and a control module. Accordingly, by using the system extensible for performing stereo matching, depth information of corrected images can be acquired in real-time without using computer systems or software programs for special purposes. Furthermore, since the system extensible for performing stereo matching can be simply realized by hardware, the system and the method of the present invention can be easily applied to actual intellectual-type robots, industrial settings, etc.

Abstract: An apparatus and method creating a ghost-free High Dynamic Range Image (HDRI) based on filtering are provided. It is possible to effectively prevent a ghost phenomenon from occurring when a single HDRI is created from a plurality of LDRIs, by defining a ghost area using a probability based on a global transfer function indicating a relationship for intensities of several frames, rather than searching for or identifying a ghost area in a single or each image.

Abstract: A method of processing frames obtained by multiple exposures may include categorizing, by at least one processor, a plurality of frames captured by multiple exposures into one of a reference frame and a target frame, compensating for global motion resulting from motion of a capturing apparatus capturing an object, and compensating for local motion resulting from motion of the object, based on a joint histogram between the reference frame and the target frame.

Abstract: Disclosed is a system and a method for rectifying stereo images, which are acquired by two cameras, in real-time by using a calibration matrix resulting from camera calibration. The system includes a coordinate generation module; a rectification coordinate generation module; a bilinear interpolation value generation module; a rectification coordinate memory; a bilinear interpolation memory; an image buffer; an rectification module; and a control module. A structure of a hardware system capable of real-time stereo rectification is provided, and the operation results have been verified by implementing a hardware device. The real-time stereo rectification system makes it possible to acquire rectified images in real-time without using a separate computer system or a software program.

Abstract: A robot cleaner includes a drive unit for driving a plurality of wheels, a camera disposed in a body for photographing an upper image extending perpendicular to a running direction, a controller for recognizing a position by using position information obtained from a recognition mark formed on a ceiling of a working area that is photographed by the camera, and controlling the drive unit by using the recognized position information to correspond to a target cleaning operation. Accordingly, since the recognition mark for the recognition of a current position of the robot cleaner is formed on a ceiling that has a few changes, the position recognition is performed with more accuracy and the traveling of the robot cleaner to the target work area and a cleaning operation is performed efficiently.

Abstract: Disclosed are a system and a method extensible for performing, in real-time, stereo snatching for calculating depth images with a result of searching for points of similarity by using images taken with two cameras. The system includes a coordinate creating module, a census transform module, a delay XOR calculation module, a stereo matching module, and a control module. Accordingly, by using the system extensible for performing stereo matching, depth information of corrected images can be acquired in real-time without using computer systems or software programs for special purposes. Furthermore, since the system extensible for performing stereo matching can be simply realized by hardware, the system and the method of the present invention can be easily applied to actual intellectual-type robots, industrial settings, etc.

Abstract: Disclosed is a system and a method for rectifying stereo images, which are acquired by two cameras, in real-time by using a calibration matrix resulting from camera calibration. The system includes a coordinate generation module; a rectification coordinate generation module; a bilinear interpolation value generation module; a rectification coordinate memory; a bilinear interpolation memory; an image buffer; an rectification module; and a control module. A structure of a hardware system capable of real-time stereo rectification is provided, and the operation results have been verified by implementing a hardware device. The real-time stereo rectification system makes it possible to acquire rectified images in real-time without using a separate computer system or a software program.

Abstract: A robot cleaner, robot cleaning system, and a method for controlling the same, using wireless communications an upwardly-looking camera for photographing an upper image perpendicular to a forward-looking direction of driving the robot cleaner; and a controller for controlling the driving unit to allow the robot cleaner to drive within a cleaning area according to a predetermined driving path, and compensating the driving path by analyzing the image photographed by the upwardly-looking camera. Optional features include second forwardly directed camera to provide a three dimensional image sensors for sensing walls or obstacles and transmission of data to external processor/controllers by a radio antenna. The inventive system can recognize the robot cleaner position more accurately by using a relatively constant upper image reducing movement error to a target position.

Abstract: A robot cleaner includes a drive unit for driving a plurality of wheels, a camera disposed in a body for photographing an upper image extending perpendicular to a running direction, a controller for recognizing a position by using position information obtained from a recognition mark formed on a ceiling of a working area that is photographed by the camera, and controlling the drive unit by using the recognized position information to correspond to a target cleaning operation. Accordingly, since the recognition mark for the recognition of a current position of the robot cleaner is formed on a ceiling that has a few changes, the position recognition is performed with more accuracy and the traveling of the robot cleaner to the target work area and a cleaning operation is performed efficiently.

Abstract: A method of reconstructing a tomogram of an X-ray apparatus. The tomogram reconstructing method includes obtaining model information and a transmission image of a subject. A tomogram is reconstructed from the transmission image using the model information. The reconstructed tomogram obtained is then displayed.

Abstract: A method of reconstructing a tomogram of an X-ray apparatus. The tomogram reconstructing method includes obtaining model information and a transmission image of a subject. A tomogram is reconstructed from the transmission image using the model information. The reconstructed tomogram obtained is then displayed.

Abstract: A mobile robot capable of recognizing its location and adjusting its direction in response to an obstacle in its way includes a running device, an obstacle detecting device for detecting the presence of an obstacle, a location recognizing device, a controlling portion, and a power supply. The location recognizing device includes a first vision camera directed toward the ceiling of a room and a first vision board. The first vision camera recognizes a base mark on the ceiling. The first vision board processes an image from the first vision camera and transmits the image data to the controlling portion. The obstacle detecting device includes a line laser for emitting a linear light beam toward the obstacle, a second vision camera for recognizing a reflective linear light beam from the obstacle, and a second vision board for processing image data captured by the second vision camera.