JP2002010105A - Method and device for eliminating noise - Google Patents

Abstract

PROBLEM TO BE SOLVED: To solve the problem that human visual characteristics do not agree, and visual deterioration in image quality is generated since noise elimination characteristics are uniform regardless of the size of motion of an image signal in a conventional noise-eliminating device. SOLUTION: The amount of pixel difference between a frame memory 103 for delaying a digital image signal for one frame period and a current frame is extracted for each frame by an absolute-value cumulative adder 105, the amount of noise elimination is successively stored into the first to third memories 106 to 108 corresponding to the time transition, motion and still pictures are judged according to data (c) to (a) that are stored into each memory, and the noise is adaptively eliminated by a noise eliminator 110.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method and an apparatus for removing noise contained in a digital video signal compressed and coded at a low bit rate by MPEG coding or the like, and more particularly to a digital video signal. The present invention relates to a method and an apparatus for removing noise from a digital video signal, which are suitable for controlling the amount of noise removal in accordance with the magnitude of signal movement.

[0002]

2. Description of the Related Art Conventionally, as a post filter for removing block distortion generated in a system for encoding a moving image signal at a low bit rate, for example, Japanese Patent Publication No. 7-1421
A block distortion removal method and apparatus proposed in JP-A No. 1 that determine generated noise based on a level difference between four pixels forming a coded block boundary and apply a smoothing process to the four pixels. Have been. A block diagram of such a prior art device is shown in FIG. Hereinafter, a conventional technique will be described with reference to FIG.

Image data input from an input unit for selecting a pixel signal level around a block boundary is a signal level of four pixels symmetrical with respect to the block boundary and arranged in a direction perpendicular to the block boundary. The image data is sequentially delayed by the input signal delay units 201, 202, and 203, and the respective outputs are input to the smoothing determination unit 204 and the smoothing operation unit 205.

[0004] The smoothing determination unit 204 determines the presence / absence of noise using the input level difference of the four pixels, and outputs an ON / OFF control signal to the smoothing calculation unit 205.

[0005] When the ON / OFF control signal is ON, the smoothing operation section 205 performs a smoothing process using the four input pixels and outputs the result. On the other hand, OF
In the case of F, the smoothing process is not performed. That is, the control of the smoothing process is performed only based on the level difference between four consecutive pixels.

[0006]

In the above-described conventional noise removal method and apparatus, the noise removal characteristics are uniform regardless of the magnitude of the motion of the video signal. For example, there is a problem that deterioration of a spatial frequency band is conspicuous in a slowly moving image having high spatial frequency sensitivity in human visual characteristics, and visual image quality deterioration occurs.

[0007] In view of the above, the present invention, in order to respond to the human visual sensitivity, according to the magnitude of the motion of the video signal,
It is an object of the present invention to provide a method and an apparatus for removing noise from a digital video signal for improving visual image quality.

[0008]

According to the noise removing method of the present invention, a pixel difference between a frame and a field of a digital video signal is extracted, and the extracted pixel difference is converted into a data over a predetermined period. And performs a moving image still image determination based on N (N is a natural number) pieces of data held for a predetermined period, and adaptively controls a noise removal amount according to the moving image still image determination result. Prepare.

[0009]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS A noise removing method according to a first aspect of the present invention extracts a pixel difference amount between any one of frames or fields of a digital video signal and calculates the extracted pixel difference amount. The data is held as data over a predetermined period, and a moving image still image determination is performed based on the N (N is a natural number) pieces of data held for the predetermined period, and the noise removal amount is adaptively determined according to the moving image still image determination result. Since the configuration for controlling is provided, it is possible to adaptively control the noise removal amount in accordance with the temporal transition between frames or fields (predetermined period) from which the pixel difference amount is extracted, thereby removing noise. Performs noise elimination control according to human visual sensitivity, and suppresses deterioration of the spatial frequency band in areas with slow human motion that has high human visual sensitivity while removing noise, thereby visually improving image quality. It is possible.

According to a second aspect of the present invention, there is provided a noise elimination method in which the moving image / still image determination in the above-described configuration classifies a video signal into an image having a large movement, an image having a small movement, and a still image. Noise removal suitable for visual sensitivity can be performed.

The magnitude of the motion is classified according to the magnitude of a predetermined threshold in the determination of a moving image or a still image, and can be adapted to human visual sensitivity by appropriately setting the threshold.

A noise removing method according to a third aspect of the present invention is characterized in that the moving image / still image determination in any one of the above-described configurations adds a configuration in which noise is not removed from an image determined to be an image having a small motion. In addition, a moving image video signal having a small motion is regarded as a quasi-still image, and noise processing can be performed separately from a complete still image having a pixel difference amount of 0. Therefore, noise processing more suitable for human visual sensitivity can be performed.

According to the noise removing method of the present invention, for example, by setting the number N of data for performing a moving image / still image determination to three, for example, a movie source can be converted to a video source for a predetermined period of time. Since the number of data to be held can be the lowest required and the number of storage means for storing data can be reduced, the cost of the apparatus can be reduced and the circuit configuration can be simplified at the same time.

[0014]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of a method and an apparatus for removing noise from a digital video signal according to the present invention will be described in detail with reference to the drawings.

FIG. 1 is a block diagram of a main part of a noise removing method apparatus according to an embodiment of the present invention. A frame memory 10 for delaying an input digital video signal by one frame period
3. A subtractor 104 that calculates a difference between a pixel displayed at the same position spatially and a value one frame before, and an absolute value of a result obtained by the subtractor 104 is calculated, and the calculation is performed over one frame period. An absolute value accumulator 105 for accumulating the results, a first memory 106 for updating and storing the absolute value accumulative addition result calculated for each frame, and a second memory for updating and storing data in the first memory 106 for each frame. Memory 107, a third memory 108 for updating and storing data of the second memory 107 for each frame, and a moving image still image for performing a moving image still image determination for each frame from the data stored in the first to third memories. A decision unit 109 for removing a noise component from an input digital video signal by using a spatial frequency band limiting filter or a smoothing filter; Constructed.

The operation will be described in detail with reference to FIG.
The digital video signal input to the input terminal 102 is input to the frame memory 103 and is delayed by one frame period, and at the same time, the subtracter 104 and the noise remover 11
Input to 0. In the subtractor 104, the digital video signal input from the input terminal 102 and the frame memory 10
3 is subtracted from the signal one frame period before, and the result is output to the absolute value accumulator 105. The absolute value accumulator 105 converts the result of the subtractor 104 into an absolute value and performs a cumulative addition process. However, since the absolute value accumulator 105 is reset to 0 by the frame pulse input from the input terminal 101 and indicating the head of the frame, the accumulated value for each frame is extracted.

The cumulative addition value extracted for each frame by the absolute value cumulative adder 105 is determined by the first
Is stored in the memory 106 as data c. On the other hand, the data b previously stored in the first memory is transferred and stored in the second memory according to the frame pulse. Similarly, the data a previously stored in the second memory is transferred and stored in the third memory according to the frame pulse. In this way, the data a, b, and c stored in the first to third memories are output to the moving image / still image determining unit 109, and the moving image / still image is determined.

The determination of a moving image or a still image is performed, for example, according to the following algorithm. The still image state is set as the initial judgment output of the moving image still image judging device. Only when all the values of a, b, and c are larger than the threshold value TH1, is it determined that the image has a large motion, and the moving image state is output. In other cases, the state of the still image is maintained. On the other hand, in order to transition from the video state to the still image state,
This is performed when any of the values a, b, and c becomes smaller than the threshold value TH2. In other cases, the moving image state up to that point is held.

The moving image / still image state determined by the above five-step algorithm is output to the noise eliminator 110, and the noise eliminator 110 determines the noise elimination amount for the input digital video signal according to the state, for example, The digital video signal from which noise is adaptively removed is output to the output terminal 111 by adaptively changing the control so as to be off in the image state and on in the moving image state.

By appropriately selecting the threshold values TH1 and TH2, noise removal suitable for human visual sensitivity can be performed. That is, an image portion having a particularly high human visual sensitivity and a small motion is set to a predetermined threshold value, and a complete still image in which all values of a, b, and c become 0 is excluded as a quasi-still image. By detecting the detected image and performing no noise removal processing on the detected image, it is possible to suppress deterioration of the spatial frequency band while performing noise removal.

Note that the above algorithm is an example,
For example, by setting three or more types of thresholds TH, it is possible to perform a moving image still image determination with less erroneous detection, and it is possible to suppress a sense of incongruity caused by frequent switching of the moving image still image determination.

Further, in the above description of the embodiment, it has been described that the cumulative addition of the difference is performed in frame units using the frame memory, and the moving image / still image determination is performed using these, but this is performed using the field memory. The calculation may be performed on a field basis, or may be performed on a frame basis by combining the values calculated for each field using a frame memory.

In the above embodiment, the case where the number of memories for storing the absolute value cumulative addition result is three has been described.
The number of memories may be any number as long as it is one or more. However, 30 movie sources whose display number per second is 24
For example, when a video signal is input such that a field is repeatedly output for conversion to a single video source, erroneous detection can be suppressed by setting the number to three or more.

[0024]

As described above, according to the present invention,
The amount of pixel difference between frames or fields of a digital video signal is extracted, and the amount of noise removal is adaptively controlled according to this temporal transition to remove noise. Removal control can be performed, and while removing noise, the effect of suppressing the deterioration of the spatial frequency band in a region where human vision sensitivity is slow and moving can be achieved, and the image quality can be visually improved. Can be

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a main part of a noise removal device according to an embodiment of the present invention.

FIG. 2 is a block diagram showing a main part of a conventional noise elimination device.

[Explanation of symbols]

 101, 102 input terminal 103 frame memory 104 subtractor 105 absolute value accumulator 106 first memory 107 second memory 108 third memory 109 moving image still image judging device 110 noise remover

Claims (8)

[Claims] 1. A method for extracting a pixel difference amount between a frame and a field of a digital video signal, holding the extracted pixel difference amount as data for a predetermined period, and storing the N (N A moving image still image determination based on (natural number) pieces of the data, and adaptively controlling a noise removal amount according to the moving image still image determination result. 2. The noise removing method according to claim 1, wherein in the moving image / still image determination, the digital video signal is classified into a high-motion image, a low-motion image, and a still image. 3. In the moving image / still image determination, the digital video signal is classified into a large-moving image, a small-moving image, and a still image. The noise removal method according to claim 1, wherein the noise removal is not performed. 4. The method according to claim 1, wherein the number N of the data applied to the moving image / still image determination is three. 5. A pixel difference amount extracting means for extracting a pixel difference amount between either a frame or a field of a digital video signal, and a pixel difference amount extracted by said pixel difference amount extracting means over a predetermined period. Storage means for storing as data, moving image still image determination means for performing moving image still image determination based on N (N is a natural number) pieces of data stored by the storage means, and determination by the moving image still image determination means Control means for adaptively controlling the amount of noise removal according to the result. 6. The apparatus according to claim 5, wherein said moving image / still image determining means classifies the digital video signal into an image having a large motion, an image having a small motion, and a still image. 7. The moving picture / still picture judging means classifies the digital video signal into an image having a large motion, an image having a small motion and a still image. The noise removing device according to claim 5, wherein the removing operation is stopped. 8. The noise removing apparatus according to claim 5, wherein the number N of the data applied to the moving image / still image determining means is three.