Video noise reduction apparatus and method using three dimensional discrete cosine transforms and noise measurement
AssigneeMatsushita Electric Corporation of America
InventorsSaiprasad V. Naimpally, Jimmy C. Tran
Life of this patent
Estimating the term…
Abstract
A noise component of a video signal is reduced by using a three dimensional discrete cosign transform (3-D DCT) to determine the time frequency spectrum of both the video signal and the noise component, subtracting the time frequency spectrum of the noise component from the time frequency spectrum of the video signal, and converting the time frequency spectrum of the modified video signal back to the spatial time domain by using an inverse 3-D DCT.
Claims22
A method of processing a first video signal having a noise component to produce a second video signal representative of the first video signal with the noise component reduced, said method comprising the steps of:
- (a) taking corresponding data samples in a plurality of frames of the first video signal to produce three dimensional (3-D) blocks of time spatial samples of the first video signal;
- (b) transforming the 3-D blocks of time spatial samples of the first video signal to produce 3-D blocks of time frequency spectrum coefficients of the first video signal;
- (c) generating, from selected segments of the 3-D blocks of time frequency spectrum coefficients of the first video signal, a 3-D block of time frequency spectrum coefficients of the noise component;
- (d) modifying the 3-D blocks of time frequency spectrum coefficients of the first video signal as a function of the 3-D block of time frequency spectrum coefficients of the noise component to produce modified 3-D blocks of time frequency spectrum coefficients which represent the second video signal;
- and
- (e) transforming the modified 3-D blocks of time frequency spectrum coefficients which represent the second video signal to produce corresponding 3-D blocks of data samples in a plurality of frames of the second video signal.
A method of processing a first video signal having a noise component to produce a second video signal representative of the first video signal with the noise component reduced, said method comprising the steps of:
- (a) taking corresponding data samples in a plurality of frames of the first video signal to produce three dimensional (3-D) blocks of time spatial samples of the first video signal;
- (b) transforming the 3-D blocks of time spatial samples of the first video signal to produce 3-D blocks of time frequency spectrum coefficients representing the first video signal;
- (c) generating, from selected segments of the 3-D blocks of time frequency spectrum coefficients of the first video signal, a 3-D block of time frequency spectrum coefficients representing the noise component of the video signal to the substantial exclusion of any other components;
- (d) modifying the 3-D blocks of time frequency spectrum coefficients of the first video signal as a function of the 3-D block of time frequency spectrum coefficients which represent the noise component to produce modified 3-D blocks of time frequency spectrum coefficients which represent the second video signal, wherein step (d) includes the step of generating modified 3-D blocks of time frequency spectrum coefficients which represent the second video signal from sums and differences of corresponding coefficients of the 3-D first video signal and the 3-D blocks of time frequency spectrum coefficients which represent the noise component;
- and
- (e) transforming the modified 3-D blocks of time frequency spectrum coefficients which represent the second video signal to produce corresponding 3-D blocks of data samples in a plurality of frames of the second video signal.
A system for processing a first video signal having a noise component to produce a second video signal representative of the first video signal with the noise component reduced, said system comprising:
- sampling means for taking corresponding data samples in a plurality of frames of the first video signal to produce three dimensional (3-D) blocks of time spatial samples of the first video signal;
- first transforming means for transforming the 3-D blocks of time spatial samples of the first video signal to produce 3-D blocks of time frequency spectrum coefficients of the first video signal;
- noise generating means for generating, from selected segments of the 3-D blocks of time frequency spectrum coefficients of the first video signal, a 3-D block of time frequency spectrum coefficients of the noise component;
- modifying means for modifying the 3-D blocks of time frequency spectrum coefficients of the first video signal as a function of the 3-D block of time frequency spectrum coefficients of the noise component to produce modified 3-D blocks of time frequency spectrum coefficients which represent the second video signal;
- and
- second transforming means for transforming the modified 3-D blocks of time frequency spectrum coefficients which represent the second video signal to produce corresponding 3-D blocks of data samples in a plurality of frames of the second video signal.
A system for processing a first video signal having a noise component to produce a second video signal representative of the first video signal with the noise component reduced, said system comprising:
- sampling means for taking corresponding data samples in a plurality of frames of the first video signal to produce three dimensional (3-D) blocks of time spatial samples representing the first video signal;
- first transforming means for transforming the 3-D blocks of the spatial samples of the first video signal to produce 3-D blocks of time frequency spectrum coefficients representing the first video signal;
- noise generating means for generating, from selected segments of the 3-D blocks of time frequency spectrum coefficients of the first video signal, a 3-D block of time frequency spectrum coefficients representing the noise component of the first video signal to the substantial exclusion of any other components;
- modifying means for modifying the 3-D blocks of time frequency spectrum coefficients of the first video signal as a function of the 3-D block of time frequency spectrum coefficients which represent the noise component to produce modified 3-D blocks of time frequency spectrum coefficients which represent the second video signal, wherein the modifying means includes means for generating modified 3-D blocks of time frequency spectrum coefficients, which represent the second video signal, using sums and differences of corresponding coefficients of the 3-D blocks of time frequency spectrum coefficients of the first video signal and the 3-D block of time frequency spectrum coefficients which represent the noise component;
- second transforming means for transforming the modified 3-D blocks of time frequency spectrum coefficients which represent the second video signal to produce corresponding 3-D blocks of data samples in a plurality of frames of the second video signal.
Dependencies are inferred from the claim text.