# US 2006/0044475 A1 · Adaptive bidirectional filtering for video noise reduction

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2006/0044475 A1
- Document ID: US20060044475A1
- Status: Published application
- Kind: A1
- Application number: 10/934,025
- Filed: 2004-09-02
- Published: 2006-03-02
- Earliest claimed: 2004-09-02
- Assignees: Samsung Electronics Co., Ltd.
- Inventors: Peng Lin; Yeong-Taeg Kim
- Web page: https://patentgrep.com/patent/US20060044475A1

## CPC

- H04N 5/21: Circuitry for suppressing or minimising disturbance, e.g. moiré or halo
- Y10S 348/91: Flicker reduction
- H04N: PICTORIAL COMMUNICATION, e.g. TELEVISION
- Y10S: TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS

## Abstract

A video noise reduction system for reducing video noise in a sequence of video frames. In the video noise reduction system, a temporal filter computes multiple temporal average values for the video frames in different temporal directions. A motion detector computes multiple motion signal values for the video frames in different temporal directions. Finally, a control unit selects one of the temporal average values based on the motion signal values as output.

## Claims

### Claim 1

A video noise reduction method for a sequence of video frames, comprising the step of:
(a) computing multiple temporal average values for the video frames in different temporal directions;
(b) computing multiple motion signal values for the video frames in different temporal directions;
(c) selecting one of the temporal average values based on the motion signal values; and
(d) outputting the selected temporal average value.

### Claim 2 (depends on 1)

The method of claim 1, wherein step (a) of computing multiple temporal average values further comprises the steps of:
computing a total average value using temporal averaging of a center frame and its forward and backward neighboring frames within a time interval;
computing a forward average value using temporal averaging of the center frame and its forward neighboring frames within the time interval; and
computing a backward average value using temporal averaging of the center frame and its backward neighboring frames within the time interval.

### Claim 3 (depends on 1)

The method of claim 1, wherein step (b) of computing multiple motion signal values further includes the steps of:
computing a total motion signal value representing the motion among a center frame and its forward and backward neighboring frames within a time interval;
computing a forward motion signal value representing the motion among the center frame and its forward neighboring frames within the time interval; and
computing a backward motion signal value representing the motion among the center frame and its backward neighboring frames within the time interval.

### Claim 4 (depends on 3)

The method of claim 3, wherein computing the total motion signal value further includes the steps of:
computing box differences over a W×H window of each pair of consecutive frames within the time interval;
selecting the maximum values of the computed box differences; and
using a nonlinear function to generate the total motion signal value from the selected maximum values.

### Claim 5 (depends on 4)

The method of claim 4, wherein the nonlinear function has the form min[T2(σ)−T1(σ), max(MAX−T1(σ),0)]/[T2 (σ)−T1(σ)], wherein MAX represents the corresponding selected maximum values, T1(σ) and T2(σ) represent lower and upper threshold values, respectively, that depend on a global noise standard deviation σ.

### Claim 6 (depends on 3)

The method of claim 3, wherein computing the forward motion signal value further includes the steps of:
computing box differences over a W×H window of each pair of consecutive frames within the time interval;
selecting the maximum values of a plurality of the computed box differences; and
using a nonlinear function to generate the forward motion signal value from the selected maximum values.

### Claim 7 (depends on 6)

The method of claim 6, wherein selecting the maximum values includes the steps of selecting the maximum values among the first half of the computed box differences.

### Claim 8 (depends on 6)

The method of claim 6, wherein the nonlinear function has the form min[T2(σ)−T1(σ), max(MAX−T1(σ),0)]/[T2(σ)−T1(σ)], wherein MAX represents the corresponding selected maximum values, T1(σ) and T2(σ) represent lower and upper threshold values, respectively, that depend on a global noise standard deviation σ.

### Claim 9 (depends on 3)

The method of claim 3, wherein computing the back motion signal value further includes the steps of:
computing box differences over a W×H window of each pair of consecutive frames within the time interval;
selecting the maximum values of a plurality of the computed box differences; and
using a nonlinear function to generate the back motion signal value from the selected maximum values.

### Claim 10 (depends on 9)

The method of claim 9, wherein selecting the maximum values includes the steps of selecting the maximum values among the second half of the computed box differences.

### Claim 11 (depends on 9)

The method of claim 9, wherein the nonlinear function has the form min[T2(σ)−T1(σ), max(MAX−T1(σ),0)]/[T2 (σ)−T1(σ)], wherein MAX represents the corresponding selected maximum values, T1(σ) and T2(σ) represent lower and upper threshold values, respectively, that depend on a global noise standard deviation σ.

### Claim 12 (depends on 2)

The method of claim 2, wherein the step of selecting one of the temporal average values based on the motion signal values further includes the steps of:
if the total motion signal value is within a first range, then selecting the total average value as the output;
otherwise, if the forward motion signal value is within a second range, then selecting the forward average value as output;
otherwise, if the backward motion signal value is within a third range, then selecting the backward average value as the output;
otherwise, selecting the current pixel value of a center frame as the output.

### Claim 13 (depends on 12)

The method of claim 12, wherein selecting one of the temporal average values as the output, FOUT, is according to the following relations:
A(i,j)=(1−MOTb(i,j))·AVGb(i,j)+MOTb(i,j)·FC(i,j);
B(i,j)=(1−MOTf(i,j))·AVGf(i,j)+MOTf(i,j)·A(i,j);
FOUT(i,j)=(1−MOT0(i,j))·AVG0(i,j)+MOT0(i,j)·B(i,j);
wherein:
AVG0 is the total average value,
AVGf is the forward average value,
AVGb is the backward average value,
MOT0 is the total motion signal value,
MOTf is the forward motion signal value,
MOTb is the backward motion signal value,
i, j are index values of a pixel location in a two-dimensional frame, and
FC is the center frame.

### Claim 14

A video noise reduction system for reducing video noise in a sequence of video frames, comprising:
temporal filter that computes multiple temporal average values for the video frames in different temporal directions;
a motion detector that computes multiple motion signal values for the video frames in different temporal directions; and
a control unit that selects one of the temporal average values based on the motion signal values as output.

### Claim 15 (depends on 14)

The system of claim 14, wherein temporal filter computes multiple temporal average values by:
computing a total average value using temporal averaging of a center frame and its forward and backward neighboring frames within a time interval;
computing a forward average value using temporal averaging of the center frame and its forward neighboring frames within the time interval; and
computing a backward average value using temporal averaging of the center frame and its backward neighboring frames within the time interval.

### Claim 16 (depends on 14)

The system of claim 14, wherein the motion detector further computes the multiple motion signal values by:
computing a total motion signal value representing the motion among a center frame and its forward and backward neighboring frames within a time interval;
computing a forward motion signal value representing the motion among the center frame and its forward neighboring frames within the time interval; and
computing a backward motion signal value representing the motion among the center frame and its backward neighboring frames within the time interval.

### Claim 17 (depends on 16)

The system of claim 16, wherein the motion detector further computes the total motion signal value by:
computing box differences over a W×H window of each pair of consecutive frames within the time interval;
selecting the maximum values of the computed box differences; and
using a nonlinear function to generate the total motion signal value from the selected maximum values.

### Claim 18 (depends on 17)

The system of claim 17, wherein the nonlinear function has the form min[T2(σ)−T1(σ), max(MAX−T1(σ),0)]/[T2(σ)−T1(σ)], wherein MAX represents the corresponding selected maximum values, T1(σ) and T2(σ) represent lower and upper threshold values, respectively, that depend on a global noise standard deviation σ.

### Claim 19 (depends on 16)

The system of claim 16, wherein the motion detector further computes the forward motion signal value by:
computing box differences over a W×H window of each pair of consecutive frames within the time interval;
selecting the maximum values of a plurality of the computed box differences; and
using a nonlinear function to generate the forward motion signal value from the selected maximum values.

### Claim 20 (depends on 19)

The system of claim 19, wherein the motion detector selects the maximum values by selecting the maximum values among the first half of the computed box differences.

### Claim 21 (depends on 19)

The system of claim 19, wherein the nonlinear function has the form min[T2(σ)−T1(σ), max(MAX−T1(σ),0)]/[T2 (σ)−T1(σ)], wherein MAX represents the corresponding selected maximum values, T1(σ) and T2(σ) represent lower and upper threshold values, respectively, that depend on a global noise standard deviation σ.

### Claim 22 (depends on 16)

The system of claim 16, wherein the motion detector further computes the back motion signal value by:
computing box differences over a W×H window of each pair of consecutive frames within the time interval;
selecting the maximum values of a plurality of the computed box differences; and
using a nonlinear function to generate the back motion signal value from the selected maximum values.

### Claim 23 (depends on 22)

The system of claim 22, wherein the motion detector further selects the maximum values by selecting the maximum values among the second half of the computed box differences.

### Claim 24 (depends on 22)

The system of claim 22, wherein the nonlinear function has the form min[T2(σ)−T1(σ), max(MAX−T1(σ),0)]/[T2 (σ)−T1(σ)], wherein MAX represents the corresponding selected maximum values, T1(σ) and T2(σ) represent lower and upper threshold values, respectively, that depend on a global noise standard deviation σ.

### Claim 25 (depends on 15)

The system of claim 15, wherein the of selecting one of the temporal average values based on the motion signal values further includes the steps of:
if the total motion signal value is within a first range, then selecting the total average value as the output;
otherwise, if the forward motion signal value is within a second range, then selecting the forward average value as output;
otherwise, if the backward motion signal value is within a third range, then selecting the backward average value as the output;
otherwise, selecting the current pixel value of a center frame as the output.

### Claim 26 (depends on 25)

The system of claim 25, wherein the control unit selects among one of the temporal average values as the output, FOUT, according to the following relations:
A(i,j)=(1−MOTb(i,j))·AVGb(i,j)+MOTb(i,j)·FC(i,j);
B(i,j)=(1−MOTf(i,j))·AVGf(i,j)+MOTf(i,j)·A(i,j);
FOUT(i,j)=(1−MOT0(i,j))·AVG0(i,j)+MOT0(i,j)·B(i,j)
wherein:
AVG0 is the total average value,
AVGf is the forward average value,
AVGb is the backward average value,
MOT0 is the total motion signal value,
MOTf is the forward motion signal value,
MOTb is the backward motion signal value,
i, j are index values of a pixel location in a two-dimensional frame, and
FC is the center frame.
