JPH0319094Y2 - - Google Patents

Description

[Detailed Description of the Invention] "Industrial Application Field" This invention relates to a noise removal circuit applied to remove impulse-like noise contained in a digital information signal.

"Background technology and its problems" For example, when converting digital video signals to digital
When recording and reproducing on a VTR, if the reproduced digital video signal contains a large level of impulse noise, this will appear as white dots in the reproduced image. A digital low pulse filter has been used to remove this noise.
In the case of a digital low pulse filter, it is difficult to remove noise without causing the waveform of the video signal to become rounded. If you try to improve the noise removal ability, the waveform of the video signal will become more rounded. Furthermore, the digital low pulse filter requires a large number of registers, multipliers, and adders, resulting in a large-scale configuration.
The drawback was slow processing speed.

``Purpose of the invention'' This invention aims to realize a noise removal circuit that can sufficiently remove impulse noise without causing signal waveform distortion, and that has a simple configuration and a high processing speed. This is the purpose.

"Summary of the invention" This invention is based on a circuit that synchronizes three successive pieces of sampling data out of continuously supplied sampling data, and compares the level of the sampling data from this synchronization circuit. 3
The present invention includes a selection circuit that outputs sampling data having an intermediate level among the sampling data, and is configured to remove impulse noise from the sampling data from the selection circuit.

"Embodiment" An embodiment in which this invention is applied to noise removal from a digital video signal will be described below.

In FIG. 1, 1, 2, and 3 indicate delay circuits, and a digital video signal is supplied to the delay circuit 1. In FIG. The digital video signal has 8 bits of data per sample, as shown in Figure 1.
The clock pulse indicated by CK has a sampling frequency synchronized with this digital video signal.

The input data, the output data of the delay circuit 1, and the output data of the delay circuit 2 are three consecutive sampling data y _k , y _k-1 , y _k-2 , and these are the level comparison circuits 4 and 5 . , 6. The level comparison circuit 4 compares the sampling data y _k and y _k-1 , and the level comparison circuit 5 compares the sampling data y k and y k-1.
y _k-1 and y _k-2 are compared, and the level comparison circuit 6 compares the sampling data y _k and y _k-2 . Each of these level comparison circuits 4, 5, and 6 determines (A>B) (B>
Generate two comparison outputs that are high level for each scene in A).

The comparison outputs of the level comparison circuits 4, 5, and 6 are supplied to a decoder 8 via a latch circuit 7. The decoder 8 determines which of the three sampled data is at an intermediate level from the comparison outputs of the level comparison circuits 4, 5, and 6, and generates a select signal based on this determination. This select signal is passed through the latch circuit 9 to the three selectors 10, 11,
12. This selector 10, 11, 1
2 is supplied with three sampling data delayed by a delay circuit 3. selector 10,1
Sampling data is taken out from either one of yk _{-1 and yk-1} through the latch circuit 13.

In one embodiment of this invention described above, FIG.
The operation when the sampling data yk _-1 is noise as shown in is explained. In the case shown in FIG. 2A, the sampling data _yk-2 is at an intermediate level, so the selection circuit 12 selects and outputs the sampling data _yk-2 . Therefore, the output data becomes as shown in FIG. 2B, and impulse-like noise can be removed.

The fact that good noise removal can be achieved by selecting intermediate level data among the three sampling data will be explained with reference to FIG.

First, let the levels of sampling data y _k and y _k-2 be L ₁ and L ₃ , respectively, and as shown in FIG. 3A,
It is assumed that the level of sampling data _yk-1 matches the level exactly between _L1 and _L3 .
The noise superimposed on this sampling data _yk-1 has a Gaussian distribution as shown in 14, and it is considered that the higher the level of noise, the lower the probability of occurrence. Therefore, when the level of the real sampling data is exactly between _L1 and _L3 , the sampling data with the actual noise superimposed becomes the sampling data _yk , as shown by diagonal lines in FIG. 3A. and y _k-2 have equal probabilities of being larger than levels L ₁ and L ₃ .

Furthermore, as shown in Figure 3B, assuming that the level of the real sampling data is closer to the level _L3 of the sampling data _yk-2 , the actual sampling data with superimposed noise is the sampling data _yk The probability that the level L is greater than ₂ is higher than that of the level L ₁ .
Furthermore, as shown in FIG. 3C, assuming that the level of the real sampling data is closer to level L ₁ , the probability that the actual sampling data with superimposed noise will be greater than level L ₃ is There is a high probability that the level will be greater than ₁ .

As can be understood from the above explanation, noise can be removed by selecting intermediate level sampling data among the three sampling data, and sampling data with values close to the true values can be obtained. It can be replaced accordingly.

Additionally, if noise is superimposed on a waveform with a rising edge as shown by the broken line in Figure 4A,
Since the intermediate level is the sampling data _yk-2 , a signal waveform after noise removal as shown in FIG. 4B can be obtained. Therefore, even if noise is removed, the rising edge of the signal waveform will not be blunted.

Note that, as shown in FIG. 5, in the case of a waveform having an extreme value, the peak of this waveform will be lowered after noise removal. However, in most cases, the three consecutive sampling data maintain monotonicity, so the case shown in FIG. 5 is rare, and the influence on the reproduced image is small and does not pose a problem.

"Effect of the invention" According to this invention, it is possible to remove impulse-like noise that is easily noticeable. Unlike noise removal using a digital low pulse filter, noise removal can be performed reliably without causing any distortion of the signal waveform, and
The configuration can be made simpler and the processing time can be made shorter.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing the configuration of an embodiment of this invention, FIGS. 2, 4, and 5 are waveform diagrams used to explain the noise removal operation according to this invention.
The figure is a schematic diagram used to explain this invention. 1, 2, 3... Delay circuit, 4, 5, 6... Level comparison circuit, 8... Decoder, 10, 11,, 1
2...Selector.

Claims (1)

[Scope of utility model registration request] A circuit that synchronizes three successive sampling data among the continuously supplied sampling data, and compares the level of the sampling data from this synchronization circuit to determine the middle of the three sampling data. a selection circuit configured to output sampling data having a level of 1.0, and a noise removal circuit configured to remove impulse noise from the sampling data from the selection circuit.