Preserving text quality in video encoding

摘要

A video encoder receives a macro-block of an image frame, and determines whether the macro-block contains text. The video encoder computes a quantization parameter for quantizing the macro-block, with the quantization parameter computed to be smaller if the macro-block is determined to contain text. The video encoder encodes the macro-block using the quantization parameter. Text quality in the encoded macro-block is preserved.

主权项

1. A method of encoding image frames, each image frame containing a plurality of macro-blocks, each of the plurality of macro blocks formed of a plurality of pixels, the method comprising:
receiving a macro-block in an image frame; determining whether the macro-block contains text; computing a quantization parameter for quantizing the macro-block, wherein the quantization parameter is computed to be smaller than a predetermined value if the macro-block is determined to contain text, wherein the determining comprises calculating a gradient for each pixel in each of a plurality of sub-blocks in the macro-block, computing, corresponding to each of the plurality of sub-blocks, an edge pixel count value representing the number of pixels with gradients greater than a first threshold value, selecting the highest of the edge pixel count values determined by the computing, classifying the macro-block as containing text if the highest of the edge pixel count values selected by the selecting has a value between a second threshold value and a third threshold value; and encoding the macro-block using the quantization parameter.