JPS58173473A - Sonar signal display processing system - Google Patents

Abstract

PURPOSE:To prevent a whitening phenomenon on recording paper, and also to improve operation accuracy of a small data, by using the geometric mean as a moving mean value of a data, and also executing logarithmic subtraction processing. CONSTITUTION:A sonar signal intensity data containing a frequency, etc. logarithmically compressed by a logarithmic amplifier 14 is processed by a moving mean circuit provided with delaying circuits 151, 152-, a multiplier 16, etc., and reduced to a moving mean value of logarithmic compression which is brought to geometric mean by arithmetical mean of a logarithmic compressing value. This geometric mean value is approximate to a noise level and has a strong spectrum, therefore, it is prevented that the mean value becomes much larger than the noise level. Also, by non-linearization by a logarithmic value, the operation is executed with high accuracy even against a weak spectrum. Subsequently, division of a logarithmically compressed signal intensity data from a delaying circuit 17 and the moving mean value is executed by being converted to logarithmic subtraction, and addition of a constant is also executed, and an operation result of positive or an operation result of negative replaced with ''0'' is outputted. In this way, a whitening phenomenon on recording paper is prevented, and also a small data can be recorded.

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a sonar signal display processing method that effectively compresses signal strength data for each frequency or direction output from a sonar device and displays the data in gradation on recording paper.
be.

In general, a loafer gram in a sonar device is a record of the analysis results of the received signal strength for each frequency as shown in Figure 1, on a piece of paper as shown in Figure 2, with frequency on the horizontal axis and time on the vertical axis. It refers to a display that continuously displays shading. The purpose is to discover the existence of spectrum 2 and its characteristics from this display surface, but since the dynamic range of the grayscale display is smaller than that of the analysis results, appropriate dynamic range compression processing is required. Conventionally, this process was performed using the following algorithm using an arithmetic moving average regarding frequency.

Here, G is a gain constant, xo is a frequency of 0, and Vn is a display output corresponding to xn. Further, K is an integer constant and is a parameter that determines the range in which the moving average is taken. FIG. 3 is an explanatory diagram of the arithmetic processing method of the above formula 1, in which X centered at the frequency n of the signal strength data string output from the frequency analyzer 3 is shown. -K to Xn+K are transmitted to delay circuits 41, 42 . ......, 42 -H Delayed by a predetermined time T, arithmetic mean value time KT centered at frequency n, data xn is divided by the above arithmetic mean value in the divider 7, gain The constant G is multiplied by a multiplier 8 to obtain an output yn. According to this method, even if the noise level varies greatly depending on the band: since each data is standardized by the average value of that band, the output yn
appears as background noise with uniform variance over the entire band. This is shown in Figure 4,
Since each data xnFpJ is normalized by the average value of surrounding data as shown in the diagram (a), a display output having uniform density distribution over the entire band as shown in the diagram (b) is obtained.

However, the above-mentioned conventional method had the following drawbacks.

(1) When there is one strong spectrum, a phenomenon of white spots occurs on both sides of the spectrum, so even if a weak spectrum is obtained near the strong spectrum (as an analysis result), it may not be displayed.

Figure 5 shows an example of this.As shown in Figure 5(a), a weak spectrum 12 is detected near a strong spectrum 11, but the moving average value for normalizing this is due to the spectrum 11. As a result, the output y of the weak spectrum 12 becomes very small and is not displayed as shown in FIG. 2(b). This is because the moving average value should originally represent the noise level in that band.
This is due to the fact that the signal contains a large signal spectrum, resulting in a value much larger than the noise level.

Therefore, even for the strong spectrum 11 itself, the corresponding moving average value is larger than the noise level, making it difficult to emphasize the original strength. ing.

(2) Output y for the normalized value 11. is yr1=Gy′n (Formula 3
) is linear as shown by . Therefore, it was not possible to clearly display weak spectra. Here some non-linear operation, for example Yn''log y.
......If you perform the conversion as shown in (Equation 4), the 6th
As shown in the figure, since the output yn is upwardly convex with respect to y'o, fluctuations in small data areas are emphasized.

(3) If the output yn is calculated by digital processing using the formula (1 n), division must be performed twice to obtain one yn, and this division is performed with a finite bit length. The accuracy was very poor when the data was small.

In order to eliminate the above-mentioned conventional drawbacks, the present invention makes use of the fact that the arithmetic mean of the logarithmic value of data is equal to the logarithm of the geometric mean, and uses the geometric mean as the moving average value to reduce the mean value to the noise level. This approximation is performed by converting division into subtraction of logarithmic values, and its purpose is to prevent white spots on recording paper as much as possible and improve calculation accuracy for small data. There is a particular thing.

The drawings will be described in detail below.

FIG. 7 is a configuration diagram of the first embodiment of the present invention, and FIG. 8 shows the flow of data processing when the operations shown in this configuration are executed by a high-speed processor (computer). First, a digital processor (13) using a frequency analyzer CFFT analyzes an input signal and outputs a signal strength data string (xn). This data string is input to the memory area M of the high-speed processor (Fig. 8 (1)). For each data item X stored in the memory area M, the logarithmic amplifier 14 calculates 101og Store in area M (Fig. 8 (2D0) Among this data string of logarithmic values, 101ogxo
-K to 10 logXn+Kt data to delay circuits 151', 15g...15zK+
Add with a delay of a predetermined time by 1+, and then multiply by 1=
n- (FIG. 8 (3))]. After that, the data 101ogx is delayed by the time KT in the delay circuit 17.
The average value zn is calculated from n and the constant B (=, l Q1
ogb) and stores the result in memory '-M (FIG. 8(4)). All negative values in this memory M are replaced with 0 (Figure 8 (S)), and multiplier 1
After multiplying by the gain number G by 8, it is output, sent to the display, and displayed on the recording paper.

Figure 9 shows the method of storing logarithms in memory in the above process.
The integer part of l Q logx is stored in the upper 8 bits of the bit memory, and the decimal part of l Q logx is stored in the lower 8 bits.

As explained above, in the first embodiment, the geometric mean is used as the moving average, so even if there is one strong spectrum, the moving average value does not deviate as much from the noise level as in the past. This effect is noticeable for data sequences that fluctuate randomly, such as noise spectra, and the conventional white spot phenomenon is almost no longer observed. Figure 10(a) is the analysis result of signal strength data for each frequency, Figure 10(b) is the display result of the conventional method for the analysis result of Figure 10(a), and Figure 10(c) is the result of the present invention. This is a result that is not displayed depending on the method.

The spectrum shown in Figure 10(a) is 21°22
．． 23.24.25 in Figures 10(b) and (C), it is clear that the spectra 23 and 25 in Figure 10(b) are not grained and are clearly in accordance with the present invention. It can be seen that the results obtained by this method are excellent. Also, by the logarithmic method shown in Figure 9,
For small data, the number of effective bits increases, so the average value (logarithm of the geometric mean) can also be determined with high accuracy. Furthermore, the logarithmic function is an upwardly convex curve, and fluctuations in a small range of data are emphasized, thereby improving the S/N for weak spectra.

In the first embodiment, the calculation was performed using a digital processor, but it can also be implemented using an analog system based on the configuration shown in FIG. FIG. 11 shows an example of the embodiment. In the figure, 31 is a frequency analyzer, 32 is a logarithmic amplifier, 33 is an averaging circuit, 34 is a delay circuit, 35 is a fixed noise circuit, 36 is a detection circuit, and 37 is an amplifier and does not require a divider circuit like the 1M1 embodiment. In the figure, a fixed bias circuit 35 adds a fixed noise B.
This is done by applying a voltage corresponding to B=1010 gb in the first embodiment. Further, the averaging circuit 33 may be an ordinary integrating circuit, that is, a low-pass filter, but it must not have a difference in gain from the delay circuit 34.

Note that the operation of the above circuit is the same as in the first embodiment.

Further, in addition to the configuration shown in FIG. 7, the circuit for taking the moving average can be implemented by using a trans-cesal filter with the same configuration as shown in FIG. 7Y.

As explained above, according to the present invention, the dynamic range of signal strength data is compressed using the geometric mean as a moving average, so even if a strong spectrum exists, the average value does not deviate significantly from the noise level, and the white It is possible to suppress the bleed-out phenomenon and display a weak spectrum close to a strong spectrum. At the same time, logarithmic compression is performed on the signal strength data, so even weak spectra can be clearly displayed. Furthermore, the display medium is not limited to gradation display on recording paper, but can also be applied to color display using a cathode ray tube.

[Brief explanation of drawings]

The drawings serve to explain the present invention. Fig. 1 is a graph showing the signal strength for the same wave number outputted from a sonar device, Fig. 2 is a graph corresponding to the graph in Fig. 1, and Fig. 3 is a conventional FIGS. 4(a) and 4(b) are explanatory diagrams of the gradation display of the noise level. 5(a) and 5(b) are illustrations of the white spot phenomenon, FIG. 6 is an illustration of the comparison of input/output characteristics, FIG. 7 is a block diagram of the first embodiment of the present invention, and FIG. 8 is an illustration of the comparison of input/output characteristics. An explanatory diagram showing the flow of data processing. Figure 9 is an explanatory diagram of the method of storing logarithms in memory.
Figure 0 (a) is a graph showing an example of signal strength data versus frequency, Figure 10 (ω is the loaf agelum according to the conventional method for the graph of Figure 10 (-), Figure 10 (c) is the graph of Figure 10 (-). 11 is a block diagram showing a second embodiment of the present invention. 1... Recording paper, 2, 11, 12, 21, 22° 23.
24.25... Spectrum, 13... Frequency analyzer, 14:... Logarithmic amplifier> 151. l 5zy...
...15zK+1.17...Delay circuit, 16.1
8...Multiplication procedure amendment (1 reference Showa! January 121, 2018, Commissioner of the Japan Patent Office, Kazuo Wakasugi, Case 1, 1986 Patent Maro No. 05119811, Title of invention, Sonar signal display processing method 8 Amendment case) Relationship with Patent Issue Address: 1-2-24J Ikejiri, Setagaya-ku, Tokyo! Name: 1ly-Affiliate Technology** Kabuto Headquarters Director C and 1 other person) Dai-Koe 4, Agent: 101! Electric (O8) 5011-
91166 Address No. 1 Building with the same name, 2-7g11 Toranomon, Minato-ku, Tokyo Name (69921) f Plf-shi Old wJ
Ill Ko 6 Subject of amendment "Detailed explanation of the invention in the specification" 6 Inclusion of amendment Attachment 9 Contents of 6 amendment, (1) "4. Niya," on page 3, line 2 of the specification Add ``by'' between ``tokoro'' and ``tokoro.'' (2) Add "T" between "ma" and "1 da" on the 4th line of page 7 of the specification, and amend the 7th line of the same page as follows. Record

Claims (1)

[Claims] Signal strength data for each frequency or direction output from the sonar device is logarithmically compressed, the logarithmically compressed data is arithmetic averaged within a predetermined range, and the average value is subtracted from the logarithmically compressed signal strength data within that range. A sonar signal display processing method characterized in that a constant is further added to the subtracted value, positive values are left as is, and when a negative value is obtained, the value is replaced with 0.