Methods, systems, and non-transitory computer readable storage media are disclosed for reconstructing three-dimensional meshes from two-dimensional images of objects with automatic coordinate system alignment. For example, the disclosed system can generate feature vectors for a plurality of images having different views of an object. The disclosed system can process the feature vectors to generate coordinate-aligned feature vectors aligned with a coordinate system associated with an image. The disclosed system can generate a combined feature vector from the feature vectors aligned to the coordinate system. Additionally, the disclosed system can then generate a three-dimensional mesh representing the object from the combined feature vector.

Disclosed herein is a method of processing a graph-based signal using a geometric primitive, comprising: specifying the geometric primitive to be used for calculating an edge weight; obtaining a parameter for each of the geometric primitive; calculating an edge weight for each of edges within the image based on the parameter; and encoding the image based on the edge weight.

Methods and apparatus for compressing image data are described along with corresponding methods and apparatus for decompressing the compressed image data. An encoder unit, which generates the compressed image data, comprises an input arranged to receive a first image and a second image, wherein the second image is twice the width and height of the first image, a prediction generator arranged to generate a prediction texture from the first image using an adaptive interpolator, a difference texture generator arranged to generate a difference texture from the prediction texture and the second image and in encoder unit arranged to encode the difference texture.

Seismic acquisition having high geophone densities is compressed based on Functional Quantization (FQ) for an infinite dimensional space. Using FQ, the entire sample path of the seismic waveform in a target function space is quantized. An efficient solution for the construction of a functional quantizer is given. It is based on Monte-Carlo simulation to circumvent the limitations of high dimensionality and avoids explicit construction of Voronoi regions to tessellate the function space of interest. The FQ architecture is then augmented with three different Vector Quantization (VQ) techniques which yield hybridized FQ strategies of 1) FQ-Classified VQ, 2) FQ-Residual/Multistage VQ and 3) FQ-Recursive VQ. Joint quantizers are obtained by replacing regular VQ codebooks in these hybrid quantizers by their FQ equivalents. Simulation results show that the FQ combined with any one of the different VQ techniques yields improved rate-distortion compared to either FQ or VQ techniques alone.

Circuity for executing operations is provided. The operations obtain pieces of coded data that is included in a bitstream and generated by coding tiles. The pieces of coded data are decoded to generate image data of the tiles. When the pieces of coded data are obtained, tile boundary independence information is further obtained, which indicates whether each of boundaries between the tiles is one of a first boundary and a second boundary. When the pieces of coded data are decoded, image data of a first tile is generated by decoding a first code string included in first coded data with reference to decoding information of an already-decoded tile when the tile boundary independence information indicates the first boundary, and by decoding the first code string without referring to the decoding information of the already-decoded tile when the tile boundary independence information indicates the second boundary.

Methods and decompression units for decompressing a selected sub-block of image element values from a compressed block of image element values. The method includes: identifying, from the compressed block of image element values, a pattern of a plurality of patterns of image element values associated with the selected sub-block; identifying, from the compressed block of image element values, one or more image element values associated with the selected sub-block; and generating the selected sub-block from the pattern and the one or more image element values associated with the selected sub-block.

The 3D structure and appearance of objects extracted from 2D images are represented in a volumetric grid containing quantized feature vectors of values representing different aspects of the appearance and shape of an object, such as local features, structures, or colors that define the object. An encoder-decoder framework applies spatial transformations directly to a latent volumetric representation of the encoded image content. The volumetric representation is quantized to substantially reduce the space required to represent the image content. The volumetric representation is also spatially disentangled, such that each voxel acts as a primitive building block and supports various manipulations, including novel view synthesis and non-rigid creative manipulations.

The present disclosure relates to a class-agnostic object segmentation system that automatically detects, segments, and selects objects within digital images irrespective of object semantic classifications. For example, the object segmentation system utilizes a class-agnostic object segmentation neural network to segment each pixel in a digital image into an object mask. Further, in response to detecting a selection request of a target object, the object segmentation system utilizes a corresponding object mask to automatically select the target object within the digital image. In some implementations, the object segmentation system utilizes a class-agnostic object segmentation neural network to detect and automatically select a partial object in the digital image in response to a target object selection request.

Methods and apparatus for compressing image data are described along with corresponding methods and apparatus for decompressing the compressed image data. An encoder unit, which generates the compressed image data, comprises an input arranged to receive a first image and a second image, wherein the second image is twice the width and height of the first image, a prediction generator arranged to generate a prediction texture from the first image using an adaptive interpolator, a difference texture generator arranged to generate a difference texture from the prediction texture and the second image and in encoder unit arranged to encode the difference texture.

A processor obtains pieces of coded data, which are included in a bitstream and generated by coding tiles, and tile boundary independence information, which indicates whether each boundary between the tiles is a first or second boundary. Image data of a first tile is generated by decoding a first code string included in first coded data with reference to decoding information of a decoded tile when the tile boundary independence information indicates the first boundary, and by decoding the first code string without referring to the decoding information when the tile boundary independence information indicates the second boundary. A bit string is added after the first code string to make a bit length of first coded data a multiple of a predetermined N bits, with N being an integer greater than or equal to 2.