摘要:
Video decoding innovations for multithreading implementations and graphics processor unit (“GPU”) implementations are described. For example, for multithreaded decoding, a decoder uses innovations in the areas of layered data structures, picture extent discovery, a picture command queue, and/or task scheduling for multithreading. Or, for a GPU implementation, a decoder uses innovations in the areas of inverse transforms, inverse quantization, fractional interpolation, intra prediction using waves, loop filtering using waves, memory usage and/or performance-adaptive loop filtering. Innovations are also described in the areas of error handling and recovery, determination of neighbor availability for operations such as context modeling and intra prediction, CABAC decoding, computation of collocated information for direct mode macroblocks in B slices, reduction of memory consumption, implementation of trick play modes, and picture dropping for quality adjustment.
摘要:
Video decoding innovations for multithreading implementations and graphics processor unit (“GPU”) implementations are described. For example, for multithreaded decoding, a decoder uses innovations in the areas of layered data structures, picture extent discovery, a picture command queue, and/or task scheduling for multithreading. Or, for a GPU implementation, a decoder uses innovations in the areas of inverse transforms, inverse quantization, fractional interpolation, intra prediction using waves, loop filtering using waves, memory usage and/or performance-adaptive loop filtering. Innovations are also described in the areas of error handling and recovery, determination of neighbor availability for operations such as context modeling and intra prediction, CABAC decoding, computation of collocated information for direct mode macroblocks in B slices, reduction of memory consumption, implementation of trick play modes, and picture dropping for quality adjustment.
摘要:
A browser-based email system has a thin client connected to a host mail server. The thin client implements a browser. The host mail server provides pages to the thin client, which can be rendered by the browser to present an email service to a user. The thin client is equipped to receive audio and video data and supports a user interface to facilitate capture of the audio or video data for inclusion in an email message. After a data stream is captured and stored locally at the client, the browser submits a request to the host mail server that contains a token in lieu of the data stream. The host mail server returns a response specifying a new page that contains the email message. The response also includes the token and indicates a location in the new page at which the audio or video data is to be rendered. Upon receiving the response, the browser inserts the data stream into the new page at the location and renders the new page. After the user has reviewed the email message, the user can send the email message, causing the browser to submit another request to the host mail server that contains both the email message and the audio or video data. Upon receiving this second request, the host mail server converts the email message and the audio or video data to a MIME message and forwards the MIME message to the intended recipient.
摘要:
A method for filtering data to be compressed by frame differencing is disclosed. In applications where large amounts of data must be either stored or transmitted, such as, for example, digital image data, various techniques are used to compress the data. One such technique is frame differencing wherein after a reference frame or key is encoded, only the differences between each frame and the immediately preceding frame in the sequence are encoded. The present invention operates as a pre-process or in conjunction with such a technique by determining which of the changed data points can be disregarded without significantly affecting, for example, the stored image, thereby reducing the amount of data which must be stored or transmitted. The determination is made in the present invention by comparing the pattern of changed data points in the vicinity of the data point under consideration with a set of pre-determined patterns.
摘要:
Methods and apparatus for information modeling, such as product modeling, are described. In one specific embodiment, a system for capturing product information includes a plurality of product trees and a plurality of configuration trees. Each configuration tree is associated with one of the plurality of product trees. At least two of the configuration trees share at least one usage configuration node, and the at least two configuration trees are associated with a same product tree. The at least two configuration trees share at least one occurrence configuration node.
摘要:
A method for filtering data to be compressed by frame differencing is disclosed. In applications where large amounts of data must be either stored or transmitted, such as, for example, digital image data, various techniques are used to compress the data. One such technique is frame differencing wherein after a reference frame or key is encoded, only the differences between each frame and the immediately preceding frame in the sequence are encoded. The present invention operates as a pre-process or in conjunction with such a technique by determining which of the changed data points can be disregarded without significantly affecting, for example, the stored image, thereby reducing the amount of data which must be stored or transmitted. The determination is made in the present invention by comparing the pattern of changed data points in the vicinity of the data point under consideration with a set of pre-determined patterns.
摘要:
A method and apparatus for managing variations in a product structure for a product. In one illustrative embodiment, a product management system for managing variations in a product structure for a product comprises a model. The model comprises a primary hierarchical organization and a number of optimized product variant structures. The primary hierarchical organization comprises a group of domain master objects. The number of optimized product variant structures comprises a group of domain configuration objects in which each domain configuration object in the group of domain configuration objects represents a configuration for a component represented by a corresponding domain master object in the group of domain master objects.
摘要:
The invention frees up memory to other uses by allocating memory to video graphics buffering only to the extent required. It also saves graphic processor bandwidth by processing, e.g. blending, only to the extent required. It does so by defining a regular grid of so-called tiles or cells in the overlay graphic data window, e.g orthogonally arrayed squares of 32-pixel sides, and allocating memory only to those requiring update, i.e. only those that are visible and subject to change between successive frames. Allocation is dynamic and requires little tagging overhead to keep track of the tiles in memory by location in the video graphic window. Processing is only of those tiles for which memory is allocated, since entirely transparent tiles within the window require no blending or processing whatsoever because they are invisible. In one implementation, only onscreen buffered graphics are tiled and the offscreen buffered graphics require full-sized memory allocation and in another offscreen buffered graphics are tiled also by clipping within variously sized buffers during successive passes through a drawing operation wherein the various sized buffers all are smaller than the entire window. New offscreen buffer-processed tiles may be swapped as they are processed for existing (visible) onscreen tiles directly, when there is less available memory, or new tiles may be processed and block-stored in memory until a wholesale swap can occur invisibly, when there is more memory available.
摘要:
A method and apparatus for lossless compression and decompression of video image data. Video data is comprised of the RGB, YUV or gray scale (color) information for each pixel of each frame of the video data. Storage of such video data can require large amounts of data. The present invention provides for lossless compression and decompression of the individual frames of video image data. Each individual frame of video image data is processed in n.times.n pixel blocks. The combination of colors (called the color list) and the number of different colors in the pixel block is then determined. If the number of colors is above a predetermined threshold, no compression occurs. If the number of colors is below a predetermined threshold, searches for prior occurrences of the combination of colors is than performed. In some instances, a color list for contiguous pixel blocks will be the same. In this instance, a pixel map representing the positioning of the colors in the pixel block is output. Further, each encountered combination of colors is stored as an entry in a color cache. The color cache is subsequently used as a color combination dictionary. In such instances of a pixel block where its color list is in the color cache, an index to the entry into the color cache is used to represent the colors found in the pixel block. The color cache need not be stored with the compressed data stream since it is reconstructed during decompression of the compressed data stream.
摘要:
Methods and apparatus for information modeling, such as product modeling, are described. In one specific embodiment, a system for capturing product information includes a product definition tree representing permitted elements of a generalized product structure, a plurality of template trees and a plurality of configuration trees. Each template tree represents a product design structure variant which represents a subset of said product definition tree. Each configuration tree represents a particular design configuration, which is constrained to implement the structure of its associated template tree.