Video encoding method with bit depth adjustment for fixed-point conversion and apparatus therefor, and video decoding method and apparatus therefor

摘要

Provided are a video encoding method of adjusting a range of encoded output data to adjust a bit depth during restoring of encoded samples, and a video decoding method of substantially preventing overflow from occurring in output data in operations of a decoding process. The video decoding method includes parsing and restoring quantized transformation coefficients in units of blocks of an image from a received bitstream, restoring transformation coefficients by performing inverse quantization on the quantized transformation coefficients, and restoring samples by performing one-dimensional (1D) inverse transformation and inverse scaling on the quantized transformation coefficients. At least one from among the transformation coefficients and the samples has a predetermined bit depth or less.

主权项

1. An apparatus configured for decoding a video, the apparatus comprising:
a receiver which is configured for obtaining transformed coefficients of a block from a bitstream; an inverse-quantization unit which is configured for generating inverse-quantized transformed coefficients by performing an inverse-quantization, wherein the performing inverse-quantization comprises performing a right bit-shifting operation based on a base-2 logarithm of a size of the block, and by performing a first clipping on the transformed coefficients, wherein the first clipping is performed after the performing the inverse-quantization and before performing an inverse transformation comprising a vertical inverse-transformation and a horizontal inverse-transformation; and an inverse transformation unit which is configured for generating sample values by performing the vertical inverse-transformation and a second clipping on the inverse-quantized transformed coefficients, and generating residual values by performing the horizontal inverse-transformation on the sample values, wherein the first clipping is performed for restricting the inverse-quantized transformed coefficients to a range of a first bitdepth, and the second clipping is performed for restricting the sample values to a range of a second bitdepth.