JPS5979687A - Device for appreciating noise of dynamic picture signal - Google Patents

Abstract

PURPOSE:To appreciate the auantity of noises included in a dynamic picture signal by detecting the movement in a screen and appreciating the quantity of noises only when the screen indicates rest. CONSTITUTION:An inputted dynamic picture signal is supplied to a dymanic vector detecting circuit 11, a frame memory 12 and a subtracting circuit 13 and a dynamic vector detecting the movement on a screen and indicating the movement is outputted. When the dynamic vector is ''0'', a gate signal generating circuit 14 discriminates static state and outputs an execution instruction to a noise appreciating circuit 15. The subtracting circuit 13 generates a frame difference by using a picture signal delayed by one frame and an input picture signal and supplies the difference value to a noise appreciating circuit 15. The circuit 15 executes appreciation in accordance with the frame difference value and the execution instruction and outputs the appreciated result to a line 1,000 every fixed period.

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a device for evaluating the amount of noise contained in a moving image signal.

There are various methods for evaluating the amount of noise contained in a moving image signal. For example, if there is a synchronized section like a television signal, and it is thought that the same amount of noise as the video signal is included in the flat part within this section, evaluate the amount of noise in this synchronized section. It is easy to measure.

However, this method does not work if there is a possibility that the video section of the television signal is left as it is in a part of the transmission system and only the synchronization section is newly assigned.
It becomes impossible to accurately evaluate the amount of noise. In fact, when a process amplifier for television signals or a time base collector used in combination with a video tape recorder (VTR) is used, the synchronization signal may be newly assigned.

Therefore, it is necessary to evaluate the amount of noise including at least the video section. There are methods for evaluating the amount of noise superimposed on the video portion. For example, if noise is superimposed on a signal with a uniform brightness level, it is also possible to subtract this brightness level and extract only the noise for evaluation. However, this method is powerless for ordinary images that do not necessarily contain a certain level, especially for moving images.

Furthermore, it is impossible to evaluate the amount of noise in a given image signal.

An object of the present invention is to provide a noise evaluation device that can evaluate the amount of noise contained in a moving image signal using video sections.

That is, the noise evaluation device of the present invention includes a detection means for detecting a motion vector indicating image motion from an input moving image signal, a storage means capable of storing approximately one frame of the moving image signal, and a motion vector that is the output of the detection means. determining means for determining whether or not the image signal corresponding to the motion vector includes motion; a subtraction means for calculating an inter-frame difference using the input moving image signal and the output of the storage means; (3) evaluation means for evaluating the amount of noise contained in the input moving image signal using an output difference signal corresponding to an image signal determined by the determination means to not contain motion;

The principle of the present invention will be explained using FIG. 1.

If it is assumed that the moving image signal contains no noise at all, there is no difference between the two screens (hereinafter referred to as frames) in the static portion, that is, the interframe difference value is zero. On the other hand, for moving parts, even if noise is not included, the frame difference value will not be zero due to the movement.

The input image signal value is X(t), the image signal value delayed by one frame time (τ) is X(t+τ), and its average power is σ
o2, and the speed of movement within the screen is represented by ■, then the average power σ- of the frame difference is expressed by the following equation.

σ'= E((x(t)-x(1+τ))2]=2σo
z(1-exp (-α-IVI)) (1) Here, EC-) is the expected value, σo2・exp (-α・1■
1) is the number of autocorrelations of the image signal when the distance between the correlated pixels is IVI, and α is the complexity of the image pattern (4
) is a constant determined by

As is clear from equation (1), in the stationary case, that is, v=
In the case of O, it is σv2-0, but in the case of V~0, it becomes σV'~0.

If the image signal contains random noise with a uniform average power of σ and ', then the average power of the input signal x (t) is σ. 2+σn2, and the average power of the frame difference at this time σ, 12 is σvl12=σv2+ σ♂ =20°"(1-exp (-α・IVI))+σ,"
It is expressed as (2).

According to equation (2), if the average power is CI, the average power of the frame difference of an image signal containing random noise is Since the term is zero, σn? becomes.

If we detect the movement in the screen in advance and calculate the average electric power (5) of the frame difference using only the motion vector (g) indicating v-0, that is, the image signal for the static part, we can calculate the average electric power (5) that is included in the original image. The average power σ♂ of the random noise that is generated can be found.

However, σ. ′) If σ♂ does not hold, a correct motion vector cannot be detected, making it difficult to accurately evaluate the amount of noise. However, in a moving image signal such as a normal television signal, σ. '/σ♂=1.000~10,000
degree, so σ. ') CI is well established.

An example of a motion vector detection method is the method described in Ninomiya's paper "Motion Correction in Interframe Coding" (paper number IE 78-6, 1978 Institute of Electronics and Communication Engineers Technical Research Report Vol. 1.78, 439). is applicable.

In this manner, the noise evaluation device of the present invention measures the amount of noise contained in an image signal using the video signal portion, and thus can accurately evaluate the amount of noise in a given image signal.

Embodiments of the present invention will be described in detail with reference to the drawings.

Figure 2 shows a block diagram of the noise evaluation circuit of the present invention (6). show.

The input moving image signals are lines 1011 and 1012.10.
13 to the motion vector detection circuit 11, frame memory 12, and subtraction circuit 13, respectively. (Kinbeku)
The motion detection circuit 11 detects movement within the screen using this input image signal and an image signal delayed by approximately one frame time supplied from the frame memory 12 via a line 1211, and generates a motion vector representing the movement. Output.

In this example, it is assumed that motion vectors are detected in units of blocks consisting of multiple pixels, as described in Ninomiya's paper mentioned above, but the same applies to the case where motion vectors are detected in units of one pixel. can.

The output motion vector is supplied to the gate signal generation circuit 14. Here, when the motion vector is zero, for example, it is stationary, it is first determined whether or not the supplied motion vector is zero, and if it is zero, that is, if it indicates stationary, it is transmitted via line 1415. and issues an execution command to the noise evaluation circuit 15.

If the motion vector is not zero, a stop command is issued.

If the motion vector is detected in a single block (7), this execution/stop command remains unchanged during the pixel time in the corresponding block. That is, when an execution command is issued, the noise evaluation circuit 15 executes noise evaluation for only the pixel time in the corresponding block.

The same applies to the case of stopping.

In the subtraction circuit 13, a frame difference is generated using the input image signal and the image signal delayed by approximately one frame supplied via the line 1213, and the difference value is expressed as a line 1.
315 to the noise evaluation circuit 15. The subtraction circuit 13 calculates the frame difference by adjusting the amount of delay in advance so that the time difference (delay) between both image signals is exactly one frame time.

In accordance with this frame difference value and the execution/stop command, the noise evaluation circuit 15 executes evaluation only when an execution command is issued, and outputs the evaluation result via the line 1000 at regular intervals (for example, once per frame). .

Next, the operation of the noise evaluation circuit 15 will be explained using FIG. 3. The frame difference signal supplied via the line 1315 undergoes conversion to square the difference value in the squaring circuit 151, and is expressed by difference power. Since this (8) differential power is not distinguished between the moving part and the stationary part, only the differential power corresponding to the stationary part is sent to the gate circuit 152 according to the run/stop command supplied via the line 1415. is output from. For a moving part, when a stop command is supplied, the output of the gate circuit 152 is set to zero, for example. The output of the gate circuit 152 is integrated by an integration circuit 153. The period of integration is, for example, one frame, and is controlled by a reset pulse supplied via the control circuit 156 through the wire 5653. This reset pulse is also simultaneously supplied to counter 155 via line 5655 and is used to reset the counter. This counter 155 calculates the number of pixels N within the execution command period among the execution/stop commands supplied via line 1415. In other words, assuming that the reset pulse supplied to the counter 155 is for each frame, the output value of this counter is determined when the number of pixels included in the still part in one frame is N.
.

At this time, the integration circuit 153 has performed integration of the pixels other than N pixels. In other words, the frame difference value included in the still part (9) is e+ (1=1.2...,
N) Then the output value of the integrating circuit 153 is 1F1e-
becomes. A division circuit 154 divides the output value of the integration circuit 153 by the output value N of the counter 155.

"+12=Σel"/N (4)-1 This is the average power of the detected noise. Since the frame difference power is only for the static part, Equation (4) includes only the noise component, so this σ, 12 matches the average power σ♂ of the frame difference of the random noise originally included in the input image signal. or have a value very close to that.

This division result is output from the noise evaluation circuit 15 via the line 1000.

The S/N ratio is usually syN ratio = 2010g, o Spp/ant (5
) is given by Here, since SPp is fixed at 0.7 V in a normal television signal, it is easy to output an 8/N ratio from the noise evaluation circuit 15.

As explained in detail above, since the present invention evaluates the noise in the video signal portion, accurate noise evaluation is possible even if the synchronization signal is replaced in the middle of the image signal supply path. No test signals are required at all, and noise evaluation can be performed for the given image signal.
When the present invention is put into practical use, its effects are extremely large.

[Brief explanation of the drawing]

Figure 1 is a block diagram explaining the invention in detail, Figures 2 and 3 are
The figure is a block diagram illustrating an embodiment according to the present invention. In the figure, 11 is a motion vector detection circuit, 12 is a frame memory, 13 is a subtraction circuit, 14 is a gate signal generation circuit, and 15
151 is a noise evaluation circuit, 151 is a square circuit, 152 is a gate circuit, 153 is an integration circuit, 154 is a division circuit, 155 is a counter, and 156 is a control circuit. Representative Patent Attorney Susumu Uban 1 (11) No. 1 Concave Chopsticks 2 Diagram

Claims (1)

[Claims] When evaluating the amount of noise contained in a moving image signal,
Detection means for detecting a motion vector indicating image movement from the motion picture signal; storage means capable of storing approximately one frame of the motion picture signal; and a motion vector that is an output of the detection means to correspond to the motion vector. determining means for determining whether or not the image signal includes motion; subtracting means for calculating an inter-frame difference using the input video signal and the output of the storage means; Noise in a moving image signal, characterized by comprising: evaluation means for evaluating the amount of noise contained in the input moving image signal using an image signal that corresponds to an image signal determined by the means to not include motion. Evaluation device.