Abstract: An image processing apparatus includes a change detection unit configured to detect a direction change, which is a temporal fluctuation in a plausible direction orthogonal to an edge determined for each pixel in an edge image indicating a high frequency component in an image that is an image processing target, and a phase change, which is a temporal fluctuation of a phase of the high frequency component according to the direction change, and a reliability estimation unit configured to estimate reliability indicating that the detected phase change is not a change caused by noise based on a variance value of the direction change per unit time.

Abstract: A camera parameter estimating device includes an obtaining unit which obtains time series frames including a plurality of subject images existing in a common field of vision shared by multiple cameras for each of the cameras, a point estimating unit which estimates the coordinates of a joint point determined in the subject image in the time-series frame for each of the subject images, a motion trajectory estimating unit which estimates the trajectory of the joint point between the time-series frames for each of the cameras, an inter-camera correspondence estimating unit which estimates a group of the trajectories between the frames obtained respectively from a pair of the cameras, and a parameter estimating unit which estimates a camera parameter for each of the cameras according to the group of trajectories.

Abstract: A skeletal information threshold setting apparatus includes: a joint information input unit that accepts an input of important joints among joints of a subject and a confidence threshold for the important joints; and a threshold setting unit that acquires a confidence threshold for each of multiple joints of the subject, including the important joints, based on the important joints and the confidence threshold for the important joints that were input, and sets the acquired confidence thresholds for the joints as thresholds to be used in making a determination regarding a skeletal estimation result for the subject.

Abstract: A reference specifying unit specifies reference values corresponding to a plurality of points of point cloud data. A residual decoding unit decodes attribute residuals of the plurality of points from encoded residual data obtained by encoding the attribute residuals, the attribute residuals of the plurality of points being residuals of attribute values of the plurality of points with respect to the reference values. An attribute value calculation unit calculates, for each of the plurality of points, an attribute value of the point from an attribute residual and a reference value.

Abstract: According to an aspect of the present invention, there is provided a learning device including: a classification unit that classifies latent variables obtained from learning data used for learning into a label feature quantity and a non-label feature quantity; a decoding unit that decodes the label feature quantity and the non-label feature quantity classified by the classification unit by using decoder parameters to generate reconstruction data; and an optimization unit that optimizes the decoder parameters to minimize a classification error between the label feature quantity and label information used for classification by using the label feature quantity, and minimize a reconstruction error by using the label feature quantity and the non-label feature quantity.

Abstract: A decoding method is a decoding method performed by a decoding device, the decoding method including steps of: acquiring, from among a plurality of slice numbers indicating a plurality of slices associated with encoded data of the entire point cloud distributed in a three-dimensional space, a first slice number indicating a first slice associated with encoded data of a point cloud of a first space that is a portion of the three-dimensional space, on the basis of a pointer indicating the first slice; and acquiring the encoded data of the point cloud of the first space on the basis of the first slice number, and decoding point cloud data of the first space from the encoded data of the point cloud of the first space.

Abstract: According to an aspect of the present invention, there is provided a learning device including: a classification unit that classifies latent variables, which are feature quantities obtained from learning data used for learning, by using a label feature quantity having label information used for classification; a decoding unit that decodes the latent variables to generate reconstruction data by using predetermined decoding parameters; and an optimization unit that optimizes the decoding parameters to minimize a classification error between the label feature quantity and the label information by using the label feature quantity.

Abstract: Provided is technology for extracting a person region from an image, that can suppress preparation costs of learning data. Included are a composited learning data generating unit that generates, from already-existing learning data that is a set of an image including a person region and a mask indicating the person region, and a background image to serve as a background of a composited image, composited learning data that is a set of a composited image and a compositing mask indicating a person region in the composited image, and a learning unit that learns model parameters using the composited learning data.

Abstract: A motion classification model learning apparatus that learns a model for early recognizing a motion is provided. A training data acquisition part acquiring training data configured with pairs of video information about a motion that can be classified into any of a plurality of categories according to characteristics of the motion and category information that is a correct label corresponding to the video information; a motion history image generation part generation a motion history image of the video information; and a model learning part learning a model that outputs a label that is the category information, with the motion history image as an input are included.

Abstract: An estimation method is an estimation method executed by an estimation device, and includes a feature estimation step of estimating a feature quantity regarding coordinates of a point cloud, a depth derivation step of deriving a depth vector of the point cloud on the basis of predetermined parameters, a shift amount estimation step of deriving a shift amount of the coordinates of the point cloud on the basis of the feature quality, a shift processing step of shifting the coordinates of the point cloud on the basis of the shift amount and the depth vector, and a plane estimation step of estimating parameters of a plane that fits the coordinates of the point clouds including the point cloud whose coordinates have been shifted.

Abstract: A subject tracking device includes a first histogram generation unit configured to generate a first histogram representing an appearance frequency of an image feature quantity of a subject region, for each segment and for each auxiliary segment of the image feature quantity, a second histogram generation unit configured to generate, for each candidate region, a second histogram representing an appearance frequency of the image feature quantity of an image of a candidate region for each segment and for each auxiliary segment of the image feature quantity, a difference derivation unit configured to compare the first histogram with the second histogram for each segment and for each auxiliary segment, and derive, for each candidate region, a total difference value of the appearance frequencies of the image feature quantities of the first histogram and the second histogram; and a region selection unit configured to select the candidate region having a minimum total difference value from among the candidate regions

Abstract: An image encoding method is an image encoding method executed by an image encoding device, the method including: a feature map generating step of generating a first feature map representing a feature of an encoding target image which is an encoding target image and a second feature map representing a feature of the encoding target image at different resolutions; a correlation map generation step of generating a correlation map representing a correlation distribution between the first and second feature maps; a contraction function generation step of generating a contraction function which is a function used for a contraction process for a predetermined image in a decoding process based on the correlation map; and an encoding step of executing an encoding process on the contraction function and outputting a result of the encoding process.

Abstract: An image processing apparatus includes a change detection unit configured to detect a phase change having a predetermined amount of change from phase changes in a subregion of an edge image having a predetermined resolution, a change amount adjustment unit configured to adjust the amount of change of the detected phase change, an adjusted image generation unit configured to generate an edge image having the predetermined resolution in which the amount of change of a motion change is adjusted based on the phase change in which the amount of change of the phase change is adjusted, and a region decision unit configured to decide a new subregion of an edge image having a new predetermined resolution based on the edge image having the predetermined resolution in which the amount of change of the motion change is not adjusted and the edge image having the predetermined resolution in which the amount of change of the motion change is adjusted.

Abstract: A data update method for replacing at least part of encoded point cloud data in which presence or absence of a point is represented by a multi-tree structure for an individual divided region of a region indicated by point cloud data without decoding the multi-tree structure includes acquiring replacement destination data, acquiring replacement source data, which is the encoded point cloud data of the individual divided region corresponding to data to be replaced, and replacing the replacement source data with the replacement destination data.

Abstract: A noise determination apparatus includes an acquisition unit and a determination unit. The acquisition unit acquires n-branch tree structure data. The n-branch tree structure data includes information on the presence or absence of a point for a divided region obtained by dividing, by an n-branch tree structure including a plurality of layers, a spatial region represented by point cloud data including information on a position of a point, and includes, for the divided region including only one point, a sign representing the point instead of information on a divided region in a layer lower than the divided region. The determination unit determines a point represented by the sign to be noise in a layer higher than a predetermined layer.

Abstract: An aspect of the present invention is a learning method for learning a model which reconstructs data and acquires reconstructed data, the learning method includes a plurality of stages, and the learning method includes first sub-processing which is based on an iterative calculation algorithm in which a Lagrange multiplier is not used, second sub-processing which is based on an iterative calculation algorithm in which the Lagrange multiplier is used, and weighting sub-processing which determines processing in each of the plurality of stages based on a weight for determining usage of the first sub-processing and the second sub-processing in each stage.

Abstract: One aspect of the present invention is a data generation method which generates data based on a predetermined estimation model, the data generation method includes a generation step of generating data estimated as a predetermined label by the estimation model and provided with the predetermined label, and generated data has at least either one of a feature close to data to which a label different from the predetermined label is imparted or a feature different from known data to which the predetermined label is imparted.

Abstract: A parameter optimization method includes extracting a feature vector using input data, acquiring a classification result of the feature vector and a class representative vector of every class serving as a classification target, and optimizing a parameter used in the extracting based on a classification error obtained using correct answer data and the classification result and a distance error between the class representative vectors such that areas of features of the classes in a feature space do not overlap each other.

Abstract: A display device which can reproduce kinematic parallax and express a high sense of realism without using image display means are provided. The display device includes an image display unit having a stripe structure having subpixels of a plurality of colors disposed so that subpixels of the same color are arranged in a first direction and enabling an observer to observe, through an aperture, an image formed by pixels, each pixel being constituted by the subpixels of a plurality of colors. The aperture has a shape that has been smoothed to reduce a number of corners in which areas of the subpixels of the plurality of colors which can be seen through the aperture are uniform, and in which a numerical aperture decreases along a second direction orthogonal to the first direction. A plurality of the apertures are provided so as not to overlap with each other.

Abstract: Disclosed is a decoding method for decoding encoded data obtained by encoding a cube that includes three-dimensional point cloud data. The decoding method includes a step of obtaining point occupation states of solids obtained after division in one direction, from the encoded data, wherein the direction of the division is determined based on correlation among the point occupation states of the solids after the division. Another decoding method and a decoding apparatus are also disclosed.