摘要:
Video image stabilization provides better performance on a generic platform for computing devices by evaluating available multimedia digital signal processing components, and selecting the available components to utilize according to a hierarchy structure for video stabilization performance for processing parts of the video stabilization. The video stabilization has improved motion vector estimation that employs refinement motion vector searching according to a pyramid block structure relationship starting from a downsampled resolution version of the video frames. The video stabilization also improves global motion transform estimation by performing a random sample consensus approach for processing the local motion vectors, and selection criteria for motion vector reliability. The video stabilization achieves the removal of hand shakiness smoothly by real-time one-pass or off-line two-pass temporal smoothing with error detection and correction.
摘要:
A video decoder is disclosed that uses metadata in order to make optimization decisions. In one embodiment, metadata is used to choose which of multiple available decoder engines should receive a video sequence. In another embodiment, the optimization decisions can be based on length and location metadata information associated with a video sequence. Using such metadata information, a decoder engine can skip start-code scanning to make the decoding process more efficient. Also based on the choice of decoder engine, it can decide whether emulation prevention byte removal shall happen together with start code scanning or not.
摘要:
Reduced latency video stabilization methods and tools generate truncated filters for use in the temporal smoothing of global motion transforms representing jittery motion in captured video. The truncated filters comprise future and past tap counts that can be different from each other and are typically less than those of a baseline filter providing a baseline of video stabilization quality. The truncated filter future tap count can be determined experimentally by comparing a smoothed global motion transform set generated by applying a baseline filter to a video segment to those generated by multiple test filter with varying future tap counts, then settings the truncated filter future tap count based on an inflection point on an error-future tap count curve. A similar approach can be used to determine the truncated filter past tap count.
摘要:
Adjustment of hardware acceleration level in a video decoder utilizing hardware acceleration is described. Errors are detected in a bitstream as it is decoded using different levels of error detection based on decoding characteristics. A statistical analysis is performed on the error values as they are detected. In one technique, if the bitstream is categorized as fitting a high error rate state in a bitstream model, then hardware acceleration is dropped. In another technique, error statistics based on run-lengths of good and bad bitstream units are kept, and compared to predetermined thresholds. If the thresholds are exceeded, the hardware acceleration level is dropped. The level is dropped in order to take advantage of superior error handing abilities of software-based decoding over hardware-accelerated decoding.
摘要:
Error detecting and protection innovations for video decoders are described. For example, in a multithreaded video decoder, a picture extent discovery (PED) task detects an error in a video bitstream which corrupts a picture. The PED task then determines any PED sub-stage which have been completed for the picture, and based on this determination, performs error-handing PED operations. In another example, an entropy decoding (ED) task checks validity on a macroblock-by-macroblock basis using a redundant buffer to avoid overflows. Additionally, error recovery innovations are described which facilitate playback of a video bit stream at an arbitrary position. For example, a video decoder chooses a picture in the bit stream after the arbitrary position at which to begin decoding based on a determination of acceptable recovery time and/or acceptable picture quality.
摘要:
A battery operated device, having a display with two or more available refresh rates, has its refresh rate selected so as to match the video frame rate of video data played back on the display. This selection is made by coordinating the resources in the device that are used to process the video from its reception through to its display.
摘要:
Reference pictures can be created to assist with video decoding. For example, a method for decoding video can comprise receiving an encoded video bit stream, determining that a reference picture is needed, and creating and inserting the reference picture into the encoded video bit stream. A method for decoding video can also comprise receiving an encoded video bit stream, performing bit stream parsing, determining that a reference picture is needed, selecting a representation level for the reference picture, and conveying data for the selected reference picture to a hardware accelerated graphics processing unit. Various video operations, such as creating reference pictures and related information, can be performed by central processing units, while other video decoding operations can be performed by graphics processing units.
摘要:
A video bit stream with pictures comprising inter-coded content can be decoded upon receiving a channel start or file seek instruction. Pictures for beginning decoding and display of the bit stream can be selected based at least in part on one or more tuning parameters that set a preference between a latency of beginning to display video and possible defects in the displayed video. In some embodiments, to implement decoding upon a channel start or file seek, one or more types of data are generated for one or more pictures. For example, picture order counts are generated for pictures after a channel start or file seek operation. As another example, a decoder generates a frame number value that triggers re-initialization of a reference picture buffer before decoding after a channel start or file seek operation.
摘要:
Reference pictures can be created to assist with video decoding. For example, a method for decoding video can comprise receiving an encoded video bit stream, determining that a reference picture is needed, and creating and inserting the reference picture into the encoded video bit stream. A method for decoding video can also comprise receiving an encoded video bit stream, performing bit stream parsing, determining that a reference picture is needed, selecting a representation level for the reference picture, and conveying data for the selected reference picture to a hardware accelerated graphics processing unit. Various video operations, such as creating reference pictures and related information, can be performed by central processing units, while other video decoding operations can be performed by graphics processing units.
摘要:
A method of extracting regions of homogeneous texture in a digital picture divides the digital picture into blocks, and for each block generates a feature vector as a function of the data moments. From the feature vectors a gradient for each block is extracted in one of two ways, either using a weighted Euclidean distance between the feature vectors or a probability mass function-based distance metric. The gradients are submitted to morphological preprocessing to remove small bumps in the gradient field. A watershed algorithm is then applied to the preprocessed gradient field to segment the gradient field into a set of spatially connected regions of homogeneous texture.