Image processing apparatus and image processing method

摘要

An image processing apparatus includes a predictive picture generator configured to generate a predictive picture for an encoding target block based on pixel values of pixels included in already-encoded blocks around the encoding target block, a predictive error generator configured to generate predictive error based on a picture expressed by the encoding target block and a weighting unit configured to apply weights to pixel values by conducting a multiplication that multiplies by a constant greater than 1 the pixel values of pixels used to generate a predictive picture for another encoding target block, the pixels being pixels from among a plurality of pixels included in predictive error generated by the predictive error generator.

主权项

1. An image processing apparatus comprising:
a processor; and a memory which stores a plurality of instructions, which when executed by the processor, cause the processor to execute; generating a predictive picture for an encoding target block based on pixel values of pixels included in already-encoded blocks around the encoding target block; generating predictive error based on a picture expressed by the encoding target block and the predictive picture; applying weights to pixel values by conducting one or both of a first multiplication that multiplies by a first constant greater than 1 the pixel values of pixels used to generate a predictive picture for another encoding target block, and a second multiplication that multiplies by a second constant less than 1 the pixel values of pixels not used to generate the predictive picture for another encoding target block, the pixels being pixels from among a plurality of pixels included in the predictive error; generating transform coefficients by applying an orthogonal transform to the predictive error; generating quantization coefficients by quantizing the transform coefficients; generating reverse quantization coefficients by reverse quantizing the quantization coefficients; generating inverse orthogonal transform data by applying an inverse orthogonal transform to the reverse quantization coefficients; reconstructing the predictive error by conducting one or both of a first division that divides by a third constant greater than 1 the pixel values of pixels used to generate the predictive picture for another encoding target block, and a second division that multiplies by a fourth constant less than 1 the pixel values of pixels not used to generate the predictive picture for another encoding target block, the pixels being pixels from among a plurality of pixels included in the inverse orthogonal transform data; generating a decoded picture based on the reconstructed predictive error and the predictive picture; and changing the value of the first constant for each pixel according to the usage frequency of pixels used to generate the predictive picture for another encoding target block from among a plurality of pixels included in the predictive error, wherein a frequency with which pixels at the lower right of the encoding target block are used is higher than a frequency with which pixels lining a right edge or pixels lining a bottom edge are used.