JPH04160383A - Smoothing circuit - Google Patents

Abstract

PURPOSE:To widen a measurement range to enable calculation at normal speed by providing a first address generation circuit for reading the received signal of RAM by azimuth direction, a second address generation circuit read from ROM and a cumulative multiplier. CONSTITUTION:RAM 2 stores a detected signal by all azimuth directions. An address generation circuit 3 is written into RAM 2 to give a reading address. ROM 4 stores the weight for an input signal X(m,n). An address generation circuit 5 gives a reading address ROM 4. A cumulative multiplier 6 gives (multiply) weight from ROM 4 to the input X(m,n) given by a tri-state buffer 1 to cumulate them. A timing generation circuit 7 transmits timing signals to the tri-state buffer 1, RAM 2 and the multiplier 6 respectively. By the above constitution, since a quantity of calculation on the circumference finding movement average for the thinned circumference is distributed, the whole quantity of calculation does not change even if a measurement range is widened and it is possible to process it with the use of a circuit having the conventional calculation speed.

Description

[Detailed description of the invention] [Industrial application field]

The present invention relates to a smoothing circuit for interpolating azimuth data in a scanning sonar that detects underwater in all directions.

[Conventional technology]

In scanning sonar, the receiving beam rotates at a constant angular velocity to detect incoming waves from all directions one after another, so as shown in Figure 2, the search is performed in a spiral pattern with the own ship at the center. This means that Incidentally, since the turning speed is high at several hundred to several thousand times in 7 seconds, a denser spiral is created. By the way, the received signal is not always obtained stably in the azimuth direction, and sometimes the signal is temporarily interrupted. Therefore, if the detection signal is displayed as it is, the displayed object will be missing, so it is necessary to interpolate the missing signal. For this purpose, for the signals S1 to S from the same direction in each station L1 to L9, for example, the average value of Sl, S7, and S3 is adopted as the signal of S, and
As the signal of 3, the average value of S2, S3, and S4 is adopted,
By taking a moving average in this way, missing signals are interpolated, and this process is called smoothing. A circuit for this smoothing is shown in FIG. A multiplier 52 multiplies the output of the adder 51 by k (<1), outputs this value Y, and inputs it to the delay circuit 53, which outputs Y-, which is delayed by one round, and The input X is added to the adder 51. In this circuit, the input X is, for example, the one shown in FIG. ,X,,X,・X l 3 is given,
If K=1/2, the output Y will be as shown in Figure 5.
Y o, Y It Y 2”' Y 13.However,
As shown in the figure. 21 is a tri-state buffer, which outputs only when the input signal X (m, n), which is a reception signal, is stored in the RAM 22 at the subsequent stage, and becomes high impedance (open) while calculating the moving average. . Note that m is the direction, and n is the circumference argument. The RAM 22 stores detection signals from all directions. 23 is an address generation circuit that provides write and read addresses to the RAM 22. 24 is a ROM that stores weights for the input signal X(m, n);
25 is an address generation circuit that provides a read address to the ROM 24. 26 is a ROM 2 input to the input X(m, n) given from the tri-state buffer 21.
(multiplication with a weight of 4), and is an accumulation multiplier that accumulates them. A timing generation circuit 27 sends timing signals to the tristate buffer 21, RAM 22, and cumulative multiplier 26, respectively. The operation of the smoothing circuit having the above configuration will be explained below. As shown in FIG. 2, the search is performed in a spiral manner with the own ship at the center, and the detected data is stored at a predetermined address in the RAM 22 via the tri-state buffer 21 in the form of an input signal X(m, n). . Input signal X(m,
When the storage of n) is completed, the tri-state buffer 21
becomes high impedance, cutting off input and output. Next, the signals 81 to 88 on the azimuth side of the circumferences L1 to L9 are read out from the RAM 22 and input to the cumulative multiplier 26, and correspondingly, 1 to 8 are also accumulated in the weights read out from the ROM 24. The multiplier 26 is manually inputted. The accumulative multiplier 26 assigns weights from the ROM 24 to each input signal S from the RAM 22 and cumulatively adds them. Here, for the detection signals S to S8, the signals S3,
As each interpolation signal S3+ to S, l of S4, S5, and S8, for example, as shown below, a weight or kll is added to the signals at six points and is cumulatively added. S3°← + ”St + kt・sy + k3・s
s + kt・sa+ks・ss +ke−Se−■S
4'← to 1・S*+kt・S3+ to +・S, + k,
・S5+ks・Se+ke・S7・・■S5“←1・
S3+kt・3 for S4+・4 for S5+・Se+ks・S
'＋＋kg・S8・・・■S6“←kl-34+2・
Ss ten ka・S, ten four・S7+ks・Se+ke・S
8...■

[Problem to be solved by the invention]

However, if you widen the measurement range to widen the search range (in other words, increase the number of turns of the spiral), in order to obtain the same smoothing effect, the number of samples used to calculate the moving average must be increased by the increased measurement range. There must be. For example, doubling the measurement range requires doubling the number of moving average samples. For this reason, expanding the measurement range increases the amount of calculations and requires a smoothing circuit with high calculation speed. In addition, even if the range is expanded, the number of rasters (concentric display scanning lines) on the display remains constant, so it is necessary to appropriately thin out the circumferences and display them, but in order to obtain a moving average for the thinned circumferences. The calculations I had made to do this were wasted. The present invention has been made to solve the above-mentioned problems, and it is an object of the present invention to provide a smoothing circuit that can be used with a circuit having normal calculation speed even when the measurement range is expanded.

[Means to solve the problem]

The smoothing circuit of the present invention is a smoothing circuit that interpolates individual data detected from all directions with respect to a data string in the azimuth direction, and includes a RAM for storing received signals from all directions, and a RAM for storing received signals from all directions. A first address generation circuit that generates a read address for reading out the stored received signals in each azimuth direction; a ROM that stores weights for a plurality of received signals read out from the RAM in each azimuth direction; R is the weight corresponding to multiple received signals read out for each azimuth direction.
, a second address generation circuit that generates a read address to be read from the OM, and an accumulation multiplier that multiplies each received signal read from the RAM by a weight read from the ROM and accumulates the results, and a measurement range. When displaying by thinning out (N-1) pieces of data from N pieces of data in the azimuth direction in order to increase the number by N times,
A predetermined timing signal is sent to the RAM, ROM, and cumulative multiplier so that the interpolation process is not performed on the data to be thinned out, but instead a part of the interpolation process is performed on the data to be displayed. The invention is characterized in that it includes a timing generation circuit that sends out.

[Effect]

If you double the range, you will have to thin out one out of every two wires, and if you triple the range, you will have to thin out two out of every three wires, but for each lap you thin out. , the amount of calculation to be performed for each cycle for which a moving average is to be calculated is distributed.For example, if the range is twice the standard range, only half of the amount of calculation that would normally have to be performed for each cycle for which a moving average is to be calculated. Do it here, and have it done in the other half when thinning out. With this method, even if the range magnification increases to N and the number of rounds increases, the calculation time for one round will be l/N compared to when not using this method, so the overall calculation The time will be the same as for the standard range. As shown in Figure 2, the standard range is the distance between each station when searching in a spiral pattern with the own ship at the center.
The range that matches the equivalent spacing of the raster on the display and can be displayed on a one-to-one basis without thinning out is called the ml range. For example, the scanning sonar search speed is 1000 times/
seconds, the speed of sound in water is approximately 1500 m, then 0.001
Since the distance that sound travels after being reflected in a second is 0.75 seconds, the interval between each search station is 0°75 m. On the other hand, if the display has 200 rasters in full range,
When the display range is 150 m, the equivalent distance between each raster station is also 150 m/200 lines = 0.75 m/line, which is the same as the above search interval.

【Example】

FIG. 1 is a control block diagram showing one embodiment of the apparatus of the present invention. Reference numeral 1 denotes a tri-state buffer, which outputs only when the input signal X (m, n), which is a received signal, is stored in the RAM 2 at the subsequent stage, and becomes high impedance (open) during calculation of the moving average. Note that m is the direction, and n is the circumference argument. R
AM2 stores detection signals from all directions. 3 is an address generation circuit that provides write and read addresses to the RAM 2. 4 is the input signal X(m,
5 is a ROM that stores the weight for RO
This is an address generation circuit that provides a read address to M4. 6 multiplies the input X (m, n) given from the tri-state buffer l with the weight from ROM2),
It is an accumulation multiplier that accumulates them. 7 is a timing generation circuit, which includes a tri-state buffer 1 and a RAM 2.
and sends respective timing signals to the cumulative multiplier 6. 8 is a RAM, which temporarily stores the cumulative result of the cumulative multiplier 6. The operation of the smoothing circuit having the above configuration will be explained below. As shown in Fig. 2, a spiral search is performed centering on the own ship, and the detected data is stored at a predetermined address in RAM 2 via a tri-state buffer l in the form of an input signal X(m, n). . When the storage of the input signal X(m, n) in the RAM 2 is completed, the tristate buffer 1 becomes high impedance and cuts off the input and output. Then R
From AM2, signals S, -S by direction, around, direction, etc.
, . The accumulative multiplier 6 inputs RO to each input signal S from the RAM 2.
Add weights from M4 and cumulatively add them. Here, conventionally, each interpolation signal s, ) to 3 of the signals S3, S4, S6, S, for the detection signal S, to S,
As llI, for example, as shown below, signals of 6 points are given weights of 1 to 6 and are cumulatively added. S 3' ←t-8+ +kx・s, 10 to 3・sa +
k4・s4+ ks・ss + ke・Se−■S
4' ←t ”St +kt・Ss + k3”s,
+, 'S5+ks 'Se +ks"S7"'
■S s"kl"S3 ten to 2's4+it3"S5 ten to 4's+++5"s7+ke"S8"'■S 6'
←t −3* + kz”s, + kff・ss +
k4・st + ks・se + ks・Ss− ■ Now, when the measurement range is tripled, if we adopt the interpolated signals S3°, S, l of the circumference L3, L8, the circumference L4, L, no S
,',S,° are thinned out, so these cycles L
4. The calculations ■ and ■ performed in L5 will be wasted. Therefore, in the present invention, calculation
, L9 is divided into three equal parts. That is, in circumference L7, Q=O Q4-Q+, ・S4+ 4・S7 4-Q In circumference L8, Q+-R Q"Q + kt・s s+ ks's a4-Q, around, 4- R Q4-Q+ is the mouth・ss+kg・S. 4-Q However, Q is the internal register of the cumulative multiplier 6, and R is the REAM
This memory corresponds to the argument m of 8. In this way, the number of times each station performs addition is two times, for a total of six times. On the other hand, to perform all the calculations with around, Q=O Q←
S. +ks"-8s+ks・8, and the number of additions is 6. In this way, if calculation ■ is divided into 3 parts around 1, L8, and L, then at the same time as around, k4・S is executed. , S
? can be stored in RAM2. At the same time as the calculation of,k,·S8,,S
, can be stored in RAM2. At the same time as the calculation of k8 and S8, S
, can be stored in RAM2. There is an advantage. Note that the calculation division is not limited to this, and may be done as follows, for example. At circumference L7, Q=O Q4-Q + k, ·S, +, ·S. Around 4-Q, in 4-R (?-3・S on Q+, 4・S7 on +) around −Q, 4-R around Q4-Q+,・S, on +,・S8 4-Q [ [Effects of the Invention] As explained above, in the present invention, the amount of calculation performed in the period for which the moving average is to be obtained is distributed to the period to be thinned out, so that even if the measurement range is expanded, the total amount of calculation does not change. Therefore, it can be processed by a circuit having conventional calculation speed.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing an embodiment of the smoothing circuit of the present invention, FIG. 2 is a diagram showing the state of exploration in scanning sonar, FIG. 3 is a block diagram showing a conventional smoothing circuit, and FIG. 5 and 5 are diagrams showing input and output signals in the smoothing circuit of FIG. 3, and FIG. 6 is a block diagram of a conventional smoothing circuit composed of digital circuits. 1... Tri-state buffer, 2.8... RAM. 3.5... Address generation circuit, 4... ROM. 6... Accumulation multiplier, 7... Timing generation circuit. Figure 2 Figure 4 Figure 5 Figure 6

Claims (1)

[Claims] (1) A smoothing circuit that interpolates individual data detected from all directions to a data string in the azimuth direction, and includes a RAM that stores received signals from all directions, and a RAM.
a first address generation circuit that generates a read address for reading out received signals stored in the RAM in each azimuth direction; a ROM that stores weights for a plurality of received signals read out from the RAM in each azimuth direction; R is the weight corresponding to the multiple received signals read out for each azimuth direction.
It is equipped with a second address generation circuit that generates a read address to be read out from the OM, an accumulation multiplier that multiplies each received signal read out from the RAM by a weight read out from the ROM, and accumulates the results, and a measurement range. When displaying by thinning out (N-1) pieces of data from N pieces of data in the azimuth direction in order to increase the number by N times,
A predetermined timing signal is sent to the RAM, ROM, and cumulative multiplier so that the interpolation process is not performed on the data to be thinned out, but instead a part of the interpolation process is performed on the data to be displayed. 1. A smoothing circuit comprising: a timing generation circuit that sends out a timing generation circuit;