Lsi Logic Corporation

C-cube Microsystems, Inc.

Feng-Ming Wang

Charles Weckel

Hervé Brelay

Kourosh Soroushian

Siok Huang Tay

A method is provided for encoding K&#x3e;1 sequentially presented video pictures. Each of the K pictures is divided into an m&#x3e;1 row&#xd7;n&#x3e;1 column array of non-overlapping coding units of equal sizes. Each coding unit occupies a respective coding unit position in the picture from which it was divided. An arbitrary, pseudo random pattern of coding units is selected for refreshing during each of the K pictures. Each pattern selected during any given one of the K pictures includes a sequence of one or more coding units of the array. In addition, the pixels of each coding unit selected for refreshing during a kpicture occupy different pixel positions than each coding unit selected for refreshing during a preceding one of the 1to (k&#x2212;1)pictures of the K pictures. Furthermore, each pixel position of a moving picture image formed from the K pictures is selected for refreshing once over the sequence of K pictures. A system for encoding includes a source of the K pictures and an intra/inter decision circuit for selecting the coding units for refreshing over the sequence of K frames. The method or system produces an encoded signal which can be stored on a storage medium. A method and system for decoding the encoded signal is also provided.

Macroblock level intrarefresh technique for encoded video

An encoder system includes an internal counter (PTS or presentation time stamp counter). The counter provides information which enables the encoder system to resynchronize itself in the event synchronization with the video input stream is lost. Preferably, the counter is implemented in software. The encoder system also include a comparator (preferably implemented in software). The comparator is able to determine if particular regions of the input video stream have been captured at an expected time according to the internal counter within a predetermined error window by comparing the actual capture time with the expected capture time. In general, if the encoder system is synchronized with the incoming video stream, each successive blanking or active region will be captured at its expected time according to the internal counter within a predetermined error window. In this event, the timing of the various modules of the encoder system (e.g., preprocessing, spatial encoding, etc.) is based on the actual timing of the capture of a particular region according to the internal counter. In the event synchronization is lost between the video input stream and the encoder system, the encoder system resynchronizes itself with the video stream. Further, encoding of previous successively captured regions is based on the expected capture time (rather than the actual capture time) until a predetermined number (e.g., four) of regions are captured at the expected time within the error window.

Growing community of inventors

Milpitas, CA, United States of America

Feng-Ming Wang