Apparatus for compensating video motions in digital televisions

Claim 3

In a method for encoding of motion video for digital television by and apparatus including a subtracting means for generating a difference signal indicative of a difference in the spatial domain between respective video informations of a digital television signal input thereto, a transforming means for transforming the difference signal generated by the subtraction means into a frequency domain signal, a quantizing means for quantizing the transformed difference signal from the transforming means to generate a quantized difference signal of a finite number, a variable length encoding means for encoding the quantized difference signal from the quantizing means into an encoded output signal having a variable number of bits according to a generation rate of the quantized difference signal, a de-quantizing means for de-quantizing the quantized difference signal from the quantizing means, an inverse transforming means for inversely transforming the de-quantized difference signal from the de-quantizing means into a signal in the spatial domain for generating a restored difference signal, an adding means, a subframe storing means for storing video information and a motion presumption/compensation means for detecting position data of moved pixels between respective video informations, calculating positional vector amounts of the moved pixels from the detected position data and motion-compensating the video information in accordance with the calculated positional vector amounts, the method of operation comprising: a) generating a difference signal indicative of a difference between video information of a current subframe of a digital television signal input at a time t, and a reference video information of a corresponding previous subframe of the digital television signal input at a time t-1 and having been motion-compensated, the current subframe and the corresponding previous subframe each being one of a plurality of adjacent subframes comprising respectively a current frame and a previous frame of the digital television signal; b) transforming the difference signal generated by the subtraction means into a frequency domain signal; c) quantizing the frequency domain signal from the transforming means to generate a quantized difference signal of a finite number; d) encoding the quantized difference signal from the quantizing means into an encoded output signal having a variable number of bits according to a generation rate of the quantized difference signal; e) de-quantizing the quantized difference signal from the quantizing means; f) inversely transforming the de-quantized difference signal from the dequantizing means into a signal in the spatial domain for generating a restored difference signal; g) adding the restored difference signal from the inverse transforming means and the reference video information of the corresponding previous subframe; h) storing video information of the corresponding previous subframe output from the adding means; and i) detecting position data of moved pixels between the video information of the corresponding previous subframe stored in the subframe storing means and the video information of the current subframe by: detecting positions of unit blocks of m × n pixels in size of the video information within a subframe; detecting positions of moved pixels of the video information by search blocks surrounding the unit blocks for those of the unit blocks not positioned at a boundary of the subframe, and detecting positions of moved pixels of the video information by limited search blocks for those of the unit blocks positioned at a boundary of the subframe, the limited search blocks surrounding the unit blocks and being smaller in size than the search blocks such that the limited search blocks do not exceed the subframe boundary; j) calculating positional vector amounts of the moved pixels from the detected position data; and k) motion-compensating the video information of the corresponding previous subframe in accordance with the calculated positional vector amounts to thereby form said reference video information.

Claim 4

In a method for encoding of motion video for digital television by and apparatus including a subtracting means for generating a difference signal indicative of a difference in the spatial domain between respective video informations of a digital television signal input thereto, a transforming means for transforming the difference signal generated by the subtraction means into a frequency domain signal, a quantizing means for quantizing the transformed difference signal from the transforming means to generate a quantized difference signal of a finite number, a variable length encoding means for encoding the quantized difference signal from the quantizing means into an encoded output signal having a variable number of bits according to a generation rate of the quantized difference signal, a de-quantizing means for de-quantizing the quantized difference signal from the quantizing means, an inverse transforming means for inversely transforming the de-quantized difference signal from the de-quantizing means into a signal in the spatial domain for generating a restored difference signal, an adding means, a subframe storing means for storing video information and a motion presumption/compensation means for detecting position data of moved pixels between respective video informations, calculating positional vector amounts of the moved pixels from the detected position data and motion-compensating the video information in accordance with the calculated positional vector amounts, the method of operation comprising: a) generating a difference signal indicative of a difference between video information of a current subframe of a digital television signal input at a time t, and a reference video information of a corresponding previous subframe of the digital television signal input at a time t-1 and having been motion-compensated, the current subframe and the corresponding previous subframe each being one of a plurality of adjacent subframes comprising respectively a current frame and a previous frame of the digital television signal; b) transforming the difference signal generated by the subtraction means into a frequency domain signal; c) quantizing the frequency domain signal from the transforming means to generate a quantized difference signal of a finite number; d) encoding the quantized difference signal from the quantizing means into an encoded output signal having a variable number of bits according to a generation rate of the quantized difference signal; e) de-quantizing the quantized difference signal from the quantizing means; f) inversely transforming the de-quantized difference signal from the dequantizing means into a signal in the spatial domain for generating a restored difference signal; g) adding the restored difference signal from the inverse transforming means and the reference video information of the corresponding previous subframe; h) storing video information of the corresponding previous subframe output from the adding means; and i) detecting position data of moved pixels between the video information of the corresponding previous subframe stored in the subframe storing means and the video information of the current subframe by: detecting positions of unit blocks of m × n pixels in size of the video information within a subframe; detecting positions of moved pixels of the video information by search blocks surrounding the unit blocks for those of the unit blocks not positioned at a boundary of the subframe, and detecting positions of moved pixels of the video information by limited search blocks for those of the unit blocks positioned at a boundary of the subframe, the limited search blocks surrounding the unit blocks and being smaller in size than the search blocks such that the limited search blocks do not exceed the subframe boundary; j) calculating positional vector amounts of the moved pixels from the detected position data; and k) motion-compensating the video information of the corresponding previous subframe in accordance with the calculated positional vector amounts to thereby form said reference video information.