JPH0374550B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to an image flicker removal method for removing flicker superimposed on an image signal such as a television video signal. Flicker can be removed by a device with this configuration.

Generally, when broadcasting off-site relay programs on television broadcasts, if it is not possible to use lighting specially prepared for television broadcasts, in areas where the commercial power frequency is 50Hz, it is recommended to use fluorescent lamps, mercury lamps, etc. When capturing an image using an electric lamp driven by a single-phase AC power supply, a beat component occurs between the field frequency of 60 Hz such as a standard television video signal, and the brightness of the captured output image flickers. So-called frizz occurs.

Regarding the removal of flicker components of image signals that occur based on such imaging targets, conventionally, a so-called noise reducer is used to remove flicker components of extremely low frequencies using a low-pass waveform with an extremely low cut-off frequency, as will be described later. As a result, not only does the structure of the removal device become exaggerated, but also, as a side effect of the noise reducer that produces the wave action described above, in the case of moving images, the image quality deteriorates due to damage to the image signal itself. There were flaws.

An object of the present invention is to eliminate the above-mentioned conventional drawbacks, and to be able to remove flicker components superimposed on an image signal using a device with a relatively simple configuration, without causing secondary image quality deterioration of the image signal itself. An object of the present invention is to provide an image flicker removal method.

That is, the image flicker removal method of the present invention detects the average value for each successive horizontal scanning period of an image signal on which flicker components having a period having an integer ratio relationship with the surface scanning period of the image signal are superimposed, and Average values for each horizontal scanning period are averaged for each unit period consisting of a predetermined plurality of horizontal scanning periods to detect the average value for each successive unit period of the image signal, and the average value for each successive unit period is detected. is integrated for a period of one surface scanning period, and detects the average value for each sequential surface scanning period of the image signal consisting of the image signal and the flicker component superimposed on the image signal, and calculates the average value for each surface scanning period. By accumulating and averaging over a period of a plurality of sequential surface scanning periods, an average value signal consisting of the respective average values of the image signal and the flicker component is formed, and the average value signal and the average value for each surface scanning period are The flicker component is detected from the average value, and the flicker component is removed from the image signal by controlling the value of the image signal according to the value of the detected flicker component. .

The present invention will be described in detail below with reference to the drawings.

First, to examine the manner in which flicker components are generated that are superimposed on the image pickup output image signal based on AC-driven lighting as described above, we will discuss how flicker components are generated that are superimposed on the image pickup output image signal based on AC-driven lighting as described above. The flickering component that occurs on television images is caused by the following mechanism.
The brightness flickers at 20Hz. In other words, in a fluorescent lamp that is lit by periodic discharge light emission at a frequency of 50 Hz, the brightness of the light emission fluctuates at a frequency of 100 Hz, so the brightness fluctuation at a frequency of 100 Hz and the field frequency of the image signal of 60 Hz However, since the television image signal is sampled at a frequency of 60 Hz, this sampling causes a flicker component to occur during sampling for encoding the image signal. It is recognized that the so-called aliasing component is generated by the same generation principle as the above, and for example, at a sampling frequency of 60Hz
As the turning point, the difference between the brightness change of 100 Hz from the turning frequency of 60 Hz and the sampling frequency of 60 Hz is 40 Hz.
Flicker with a fundamental frequency of 20 Hz as the difference in frequency between the component with a sampling frequency of 60 Hz folded back at the 100 Hz change in brightness and the image signal component with a baseband of 60 Hz. components will be generated.

In other words, the image signal waveform obtained by imaging a white subject with uniform brightness illuminated by a discharge lamp such as a fluorescent lamp driven by a single-phase AC power supply with a frequency of 50 Hz using a television camera is converted into a field period. When observed with an oscilloscope using synchronized sweeps, as shown in Figure 1, the image signal waveform becomes a superimposed sinusoidal level change.
Level changes of approximately the same waveform appear repeatedly every eight field periods, which corresponds to the least common multiple of the field period length of Hz.

In order to remove the flicker component generated in such a manner and superimposed on the image signal without causing any damage to the image signal itself, it is first necessary to extract only the flicker component. In this method, the flicker component is extracted based on the following operating principle.

Now, let G be the image signal obtained by imaging under flicker-free illumination, let F be the alternating current component of flicker, and let F = 1 + F be the illumination with flicker. The image signal T actually obtained is T=S×(1+F). Since each term in the above equation is a function of time, the following subscripts are attached to the symbol of each term to represent the time relationship of each function.
That is, pixel numbers i are assigned to successive pixels in one line, line numbers j are assigned to successive lines in one field, and field numbers k are assigned to successive fields. Therefore, the image signal Ti,
j, k represents the value T of the i-th pixel in the j-th line of the k-th field.

However, the brightness level change component that occurs due to flicker has an extremely low fundamental frequency of 100Hz, and is a level change component that hardly changes in level over several lines to 10 lines. , if the maximum number of pixels in one line is I, then the average value of the pixel signal T in one line is Σ 0≦i≦I T/I, and each of the upper and lower M The average value of the true pixel signal S over the line is (Σ 0≦i≦I Σ 0≦i≦I Σ J−M≦j≦J+M S _i,j,k )1/I×(2M+1)=S ^l
,M _J,k . Note that S ^l,M _J,k on the right side of the above equation is an abbreviation for the left side, and M is preferably set to about 5 to 10. Therefore, the average value of the image signal T including the flicker component in such a unit of 2M+1 line period is: 1/I×(2M+1)(Σ 0≦i≦I Σ J−M≦j≦J+M T) = { Σ 0≦i≦I Σ 0≦i≦I Σ J−M≦j≦J+M S _i,j,k・(1+F _i,j,k )}1/1
×(2M+1) = (1+F _i,j,k )・S ^l,M _J,k (1) When the average value of the image signal T according to equation (1) is expressed using the above abbreviation symbol, (1+F _{i, j,k} )・S ^l,M _J,k ≒T ^l,M _J,k (1)′ On the other hand, since the content of the image signal itself generally does not change significantly between successive fields,
There is no large change in the average value of the image signal T between successive fields. Therefore, as described above with reference to FIG. 1, if the average value of the image signal T over the 8-field period in which the flicker AC component waveform goes around is calculated for the k-th field, then 1/3 Σ K≦k≦K+2 T ^l,M _J,k =Σ K≦k≦K+2 (1+F _i,J,k )・S ^l,M _J,k ×1/3 ≒1/3S ^l,M _J,k・Σ K≦k≦K+2 (1+F _i,j,k ) (2). Therefore, as mentioned above, the flicker AC component waveform appears repeatedly in almost the same waveform every three fields, so ΣF _i,j,k = 0 for every three fields, so 1/3 Σ K≦k≦ K+2 T ^l,M _J,k = 1/3 S ^l,M _J,k・Σ K≦k≦K+2 (1) = S ^l,M _J,k (3).

Note that in the above equation, Σ K≦k≦K+2 F _i,j,k =0, but this condition does not necessarily hold if the flicker AC component waveform is not a sine waveform.
However, even if the flicker AC component waveform is not a sinusoidal waveform, the average value ΣF of the flicker AC component becomes equal for every three field periods based on the repetition of the same waveform, so that value can be calculated as the DC component. By renormalizing into Σ(1) of the above equation (3), the above equation (3) can be regarded as a general equation that is unrelated to the flicker AC component waveform.

Therefore, the average value of the net image signal S is obtained by the above equation (3), and therefore, the second term on the left side of the above equation (1)', which represents the average value of the image signal T including the flicker component, Since S ^l,M _J,k can be determined, the value of the flicker component can be determined from the value S ^l,M _J,k and the value of T ^l,M _J,k on the right side. In other words, the above equation (1)′
From (1+F _i,j,k )=T ^l,M / _J,k /S ^l,M / _J,k (4) Therefore, F _i,j,k = T ^l,M / _J,k − S ^l,M / _Jk / S ^l,M / _J,k (5), and the flicker component can be easily extracted. As is clear from the above analysis, the flicker component F _i,j,k has a substantially constant value within one line, so it is sufficient to find its value for each line.

Once the flicker component is determined as described above, it is extremely easy to determine the value of the net image signal S from the value of the image signal T including the flicker component, and S _i,j,k = T _{i,j ,k} /(1+F _i,j,k ) (6). Therefore, if the purpose is to remove only the flicker component superimposed on the image signal, it is sufficient to obtain the value of the above equation (4), and by transforming the above equation (6), S _{i ,j,k} = S ^l,M / _J,k / T ^l,M / _J,k・T _i,j,k (7) Then, the above equation (7) is the final net image required. It will represent a signal.

Next, FIG. 2 shows a configuration example of an image flicker removing apparatus according to the present invention, which obtains an image signal expressed by the above equation (7) based on the above-described operating principle. In the illustrated configuration example, descriptions of conventional pre-filters, interpolation filters, etc., which are essential to this type of device, are omitted, and only parts directly related to the implementation of the present invention are shown. In the illustrated configuration example, the input image signal is first converted into an analog signal.
After converting it into a digital image signal by a digital converter 1, it is led to a subtracter 2 to subtract a synchronization signal component below the pedestal level, and then the average value of the image signal is determined for each successive line by a line averager 2. Next, the average value of the image signal is supplied to the ±M line moving averager 4 to determine the average value of the image signal in _the range of ±M lines that sequentially move. The data is switched to cyclically stored in the four memories 5A to 5D. It is assumed that the line averager 3 and the ±M line moving averager 4 are driven by a line drive signal HD and operated line by line. Furthermore, each of the four memories 5A to 5D has a memory capacity for one field, that is, 256 words corresponding to 256 lines of one field, and therefore one word.
It has a memory capacity of about 3000 bits (12 bits), and the stored data is updated every 4 fields.

The image data A to D in field units read from the four field memories 5A to 5D that perform such circular storage are input to AND gates 6A to 6D, respectively.
and ground the other input terminals of those AND gates in sequence through switch _S2 . Therefore, those four AND gates 6A to 6D
From then on, one input is grounded and closed in sequence in the field period, and the other three except the one that is closed are
The image data of the field period is taken out cyclically, integrated by an integrator 7, multiplied by a coefficient 1/3 by a multiplier 8 and averaged to 1/3, and then
It is led to the divider 9. Among the image data A to D of the sequential fields from the memories 5A to 5D, the image data of three field periods before the average image data is sequentially sent to the divider ₉ as a divisor via the switch S3. Since the image data from three field periods before is substantially equal to the image data from the current field period, the divider 9 outputs S ^l,M _J in the final equation (7) above. _,k ／
The values of T ^l,M _J,k are obtained. Therefore, the value of this ratio is led to the multiplier ₁₀ via the switch S4, and the image signal component higher than the petestal level supplied from the subtracter 2, that is, T _i,j in the final equation (7) _,k , the net image signal S _i,j,k can be obtained.Adding the synchronization signal component below the pedestal level that was previously removed by the adder 11, the digital-to-analog converter In step 12, the signal is restored to an analog signal and then extracted as a net output image signal with flicker components removed.

Note that it is preferable that the above-described calculation processing of the image average value data be performed during each blanking period to avoid disturbing the image content. In addition, all calculations from point a to point b in the circuit configuration shown in the second diagram can be performed at a calculation speed of once per line period, so even if a so-called microcomputer is used, the processing can be performed satisfactorily. You can, but
Since the arithmetic processing itself is extremely simple, even if the hardware is configured as shown in the figure, the device can be configured relatively easily.

Note that if the image content has extremely low correlation between fields, such as when changing the imaging scene, there is a possibility that a malfunction will occur in flicker component detection using the above-mentioned calculation process. Switch _S4 is switched to the "1" side as shown in the figure in response to a scene change, etc., and the original image signal T _i,j,k from which the above-mentioned flicker removal arithmetic processing has been excluded is used as an output image signal in its original form. Alternatively, the degree of flicker correction is reduced to obtain an output image signal. In addition, when switching such a switch, it is necessary to determine whether or not the imaging scene has been changed, since as mentioned above, flicker is caused by repeated level changes of almost the same waveform in three field periods. By finding the absolute value of the difference between signals that differ by 3 fields, such as a signal with a 2-field delay and a 5-field delay, or a signal with a 1-field delay and a 4-field delay, the effect of flicker can be canceled out, and this absolute value can be It is preferable that when a predetermined value is exceeded, it is determined that there is a change in the imaging scene. Furthermore, in reality, even if a malfunction occurs in flicker removal when the image content is switched, the image signal level will only deviate from the normal value for one to two field periods, and significant image quality deterioration such as image blurring will occur. Therefore, it is recognized that it is virtually unnoticeable. Furthermore, although the amplitude of the flickering AC component superimposed on the image signal due to fluorescent lamp illumination changes depending on the situation, it is not that large, so the second
If a limiter function is provided to the multiplier 10 in the illustrated configuration to suppress unnecessary large level fluctuations, the above-described malfunction problem of flicker removal can be practically solved.

Next, FIG. 3 shows an example of the configuration of the flicker removing apparatus according to the present invention in which the configuration shown in FIG. 2 is changed to a form that is easy to manufacture using hardware. Compared to the configuration shown in the second figure, the illustrated configuration consists of switches _S2 and AND gates 6A to 6D instead of the memories 5A to 5D being replaced by three 263-bit delays 13 to 15, which are essentially memory elements. The configuration is substantially the same as that shown in FIG. 2, except that the circulation selection system is simplified, the line averager 3 is replaced with a one-line period integrator 3', and the above-mentioned amplitude limiter 16 is added.

As is clear from the above description, according to the present invention, a flicker component superimposed on an image signal obtained by imaging under fluorescent lamp illumination is removed by using a device with an extremely simple configuration, and the image signal itself does not have any side effects. Flicker can be easily removed without any subsequent deterioration in image quality, and since the method of the present invention does not utilize inter-frame correlation of pixels, which has been attempted in the past, flickering can be easily removed when the image content moves. However, there is no possibility that significant image quality deterioration will occur due to the arithmetic processing for flicker removal.

[Brief explanation of drawings]

FIG. 1 is a waveform diagram showing an example of the aspect of flicker superimposition on an image signal, FIG. 2 is a block diagram showing an example of the configuration of an image flicker removal device according to the method of the present invention, and FIG. 3 similarly shows another example of the configuration. It is a block diagram. 1...Analog-digital converter, 2...Subtractor, 3...Line averager, 3'...1 line period integrator, 4...±M line moving averager, 5A to 5D...Memory, 6A to 6D...And gate, 7...Integrator,
8, 10...multiplier, 9...divider, 11...adder,
12...Digital-to-analog converter, 13-15
...Delay, 16...Amplitude limiter.

Claims (1)

[Claims] 1. Detect the average value for each sequential horizontal scanning period of an image signal on which flicker components with a period having an integer ratio relationship with the surface scanning period of the image signal are superimposed, and calculate the average value for each sequential horizontal scanning period by a predetermined number of times. The average value of the image signal is detected for each unit period consisting of a horizontal scanning period of Detecting an average value for each sequential surface scanning period of the image signal consisting of a signal and the flicker component superimposed on the image signal, and applying the average value for each surface scanning period to a period of a plurality of sequential surface scanning periods. By accumulating and averaging the image signal and the flicker component, an average value signal is formed from the respective average values of the image signal and the flicker component, and the flicker component is detected from the average value signal and the average value for each surface scanning period. An image flicker removal method characterized in that the flicker component is removed from the image signal by controlling the value of the image signal according to the value of the detected flicker component.