A method and device for decoding a point cloud using octree partitioning and a predictive tree include obtaining the point cloud. A bounding box of the point cloud is determined. Octree nodes are generated by partitioning the bounding box using octree partitioning. The predictive tree is generated for points in at least one octree node of the octree nodes. A transform is applied to the predictive tree. The points in the at least one octree node are decoded using the predictive tree.

Provided is a method for decoding a video signal based on a reduced transform, which includes: checking whether a transform skip is applied to a current block; obtaining a transform index indicating a transform kernel of the current block from the video signal when the transform skip is not applied to the current block; determining a region where a primary transform is applied to the current block based on the transform kernel indicated by the transform index and a size of the current block; and performing an inverse primary transform on the region to which the primary transform is applied by using the transform kernel indicated by the transform index.

Provided is a method for decoding a video signal based on a reduced transform, which includes: checking whether a transform skip is applied to a current block; obtaining a transform index indicating a transform kernel of the current block from the video signal when the transform skip is not applied to the current block; determining a region where a primary transform is applied to the current block based on the transform kernel indicated by the transform index and a size of the current block; and performing an inverse primary transform on the region to which the primary transform is applied by using the transform kernel indicated by the transform index.

In one embodiment, an apparatus comprises a memory and a processor. The memory is to store visual data associated with a visual representation captured by one or more sensors. The processor is to: obtain the visual data associated with the visual representation captured by the one or more sensors, wherein the visual data comprises uncompressed visual data or compressed visual data; process the visual data using a convolutional neural network (CNN), wherein the CNN comprises a plurality of layers, wherein the plurality of layers comprises a plurality of filters, and wherein the plurality of filters comprises one or more pixel-domain filters to perform processing associated with uncompressed data and one or more compressed-domain filters to perform processing associated with compressed data; and classify the visual data based on an output of the CNN.

A method for encoding a digital hologram represented by values associated respectively with pixels in a plane defining the digital hologram includes forming matrix blocks associated respectively with regions composed of contiguous pixels, each matrix block containing elements determined as a function of the values of the pixels in the region associated with the respective matrix block, applying to each of the matrix blocks a space-frequency transformation to obtain, for each matrix block, a set of coefficients respectively corresponding to different two-dimensional spatial frequencies within the respective matrix block, constructing two-dimensional structures each including coefficients from sets of coefficients and associated with two-dimensional spatial frequencies meeting a criterion that is dependent on the two-respective dimensional structure, and encoding the constructed two-dimensional structures.

A method and device for processing ultrasonic data from an industrial inspection device. Ultrasonic reflections from a high-resolution phased array are sampled at high-frequency to create a large volume of data. The data are converted to I/Q data and compressed to a manageable size. An external computer can beamform and render an image without restoring the compressed data to raw form. This method may be used for in-line inspection, downhole inspection, or ultrasonic testing.

A method and device for processing ultrasonic data from an industrial inspection device. Ultrasonic reflections from a high-resolution phased array are sampled at high-frequency to create a large volume of data. The data are converted to I/Q data and compressed to a manageable size. An external computer can beamform and render an image without restoring the compressed data to raw form. This method may be used for in-line inspection, downhole inspection, or ultrasonic testing.

An image encoding device includes a prediction unit configured to generate prediction errors being a difference between a predicted image obtained by prediction processing for an input image and the input image, a first transform unit configured to generate first transform coefficients by performing orthogonal transform on the prediction errors, a second transform unit configured to generate second transform coefficients by performing LFNST processing on the first transform coefficients, a quantization unit configured to generate quantization coefficients by performing quantization processing on the second transform coefficients, and an encoding unit configured to encode the quantization coefficients, wherein the encoding unit encodes information indicating whether a range of possible values at least taken by the second transform coefficients is to be a range determined based on a bit depth or a fixed range.

A video decoding method according to the present invention may comprise: a step for determining whether to divide a current block into a plurality of sub-blocks; a step for determining an intra prediction mode for the current block; and a step for performing intra prediction for each sub-block on the basis of the intra prediction mode, when the current block is divided into the plurality of sub-blocks.

A video decoding method according to the present invention may comprise: a step for determining whether to divide a current block into a plurality of sub-blocks; a step for determining an intra prediction mode for the current block; and a step for performing intra prediction for each sub-block on the basis of the intra prediction mode, when the current block is divided into the plurality of sub-blocks.