Video encoding method and video encoding device

摘要

A video encoding method is provided, which includes: deciding a first search window of a first macroblock in a reference image, and deciding a second search window of a second macroblock in the reference image, wherein the first search window is at least partially overlapped with the second search window; reading the first search window and the second search window from a memory and executing a first encoding stage having motion estimation and binarization encoding; storing an encoding result of the first encoding stage in the memory; reading the encoding result from the memory and executing a second encoding stage. The second encoding stage has a context entropy coding, and an encoding order of the context entropy coding is different from that of the motion estimation, and the first encoding stage is executed in parallel with the second encoding stage. Accordingly, a bandwidth demand of the memory is decreased.

主权项

1. A video encoding method, for a video encoding device, the video encoding method comprising:
deciding a first search window in a first reference image, with respect to a first macroblock of a current image, and deciding a second search window in the first reference image, with respect to a second macroblock of the current image, wherein the first search window is at least partially overlapped with the second search window; reading pixel values within the first search window and the second search window in the first reference image from a memory, and executing a first encoding stage on the first macroblock and the second macroblock of the current image, so as to generate a first encoding result with respect to the first macroblock and a second encoding result with respect to the second macroblock, wherein the first encoding stage comprises a motion estimation and a binarization encoding; executing a second encoding stage on a macroblock of a previous image, wherein executing the second encoding stage on the macroblock of the previous image comprises executing a context entropy coding on a third encoding result with respect to the macroblock of the previous image which is generated by executing the first encoding stage on the macroblock of the previous image, and the context entropy coding of the second encoding stage on the macroblock of the previous image is executed in parallel with the motion estimation of the first encoding stage on the first macroblock of the current image, wherein the binarization encoding and the context entropy coding are implemented on different encoder, wherein an encoding sequence according to which the macroblocks of the current image are processed in the context entropy coding of the second encoding stage is different from an encoding sequence according to which the macroblocks of the current image are processed in the motion estimation of the first encoding stage.