JPS63177066A - Space filter - Google Patents

Abstract

PURPOSE:To detect a speed with high accuracy, in taking out a signal from each light detection element, by applying a predetermined window function. CONSTITUTION:For example, the real image of a road surface 9 is formed on the light detection surface of a light detection element group (detector) 1 by a lens 8. Therefore, the shape of the road surface 9 is sampled in a form of brightness change by the elements 1(1)-1(N) of the detector 1. When these elements 1(1)-1(N) are arranged at a definite pitch P, a delay element is present between the mutually adjacent elements. That is, delay becomes a distance in a space filter. In these corresponding relations, at the time of movement at a constant speed, the image of one point S on the road surface 9 becomes S' on the surface of the detector 1. Therefore, the image S' of the point S successively traverses the elements 1(1)-1(N) of the detector 1 at a definite time interval and the signal due to said image S' is successively obtained from each of terminals X1-XN. By this method, a speed can be detected with high accuracy.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a spatial filter for non-contact detection of the speed of a moving object, and particularly to a spatial filter suitable for a speed detection device for an automobile.

[Conventional technology]

In recent years, it has become desirable to accurately detect the ground speed of automobiles and other vehicles for purposes such as anti-skid control, and for this reason, speed detection devices using spatial filters (also called spatial filter detectors) have become practical. It has become more and more popular.

By the way, as a conventional example of this spatial filter, for example, there is one disclosed in Japanese Patent Application Laid-Open No. 53-52473. In other words, the spatial filter according to this conventional example is
As shown in FIG. 12, a plurality of photodetecting elements 1 (1), 1 (2
), ......1 (N), and the photodetection outputs of these elements 1 (1) ~ are sent to the differential amplifier 2.
The output is a signal containing frequency components related to speed.

In FIG. 12, l is a group of photodetecting elements constituting a spatial filter, and although not shown, an image of the road surface on which a car or the like is traveling is focused on the photodetecting surface by an appropriate optical system. At this time, as the automobile or the like travels, the image of the road surface moves along the arrangement direction of each photodetector element 1 (1) of the photodetector element group 1. It is arranged like this.

By the way, the signal thus obtained at the output of the differential amplifier 2 does not necessarily contain only purely necessary signal components, but also contains various components.

Therefore, a variable BPF (band pass filter) 3 is provided after the differential amplifier 2 to extract only the desired signal component, but at this time, the frequency of the desired signal to be extracted is Since it changes depending on the speed to be detected, is there a tracking side for tracking and controlling the center frequency of this BPF 3 with respect to the speed? It is customary to provide an Im circuit 4.

The signal taken out as the output of the BPF 3 is processed by the output circuit 5, and the center frequency is estimated and output as speed information. Note that these variable BPF 3, continuous casting control circuit 4, center frequency detection circuit 5, etc. require relatively complex and advanced technology.

By the way, such conventional examples are based on the idea of compensating for the poor frequency characteristics of the spatial filter with a subsequent circuit, and little consideration has been given to improving the frequency characteristics of the spatial filter itself.

Note that Japanese Patent Application Laid-Open No. 154220/1984 may be cited as another related device of this type.

[Problem that the invention seeks to solve]

In the above conventional technology, the characteristics of the spatial filter itself are not very good as shown in FIG. As B increases, the convergence B of the main rope narrows accordingly, but the gain difference at the side ropes does not change much, and in particular the ratio with the first side rope remains almost constant regardless of N. be. Therefore, even if the number N of elements is increased somewhat, the ability to suppress other frequency components does not increase much, and for this reason, it can be said that the spatial filter shown in FIG. 12 has poor performance as a filter.

Furthermore, in such applications, the peculiarities of the spectral distribution of the input signal exacerbate the drawbacks of conventional spatial filters. In other words, in a normal spatial filter, since the dimensions of the photodetecting element are finite, it wanders as a kind of low-pass filter with respect to spatial frequencies. In addition, due to the blurring distortion of the optical system or the characteristics of the ground (Java road, etc.), the lower spectral components become larger, and sometimes the first
Even if the suppression characteristics for the spatial filter shown in FIG. 3 are included, the target signal component may even become relatively weak.

For example, in a special situation, if a periodic pattern is drawn with highly reflective white paint on a road surface with low reflectance, such as asphalt, the signal component of that period will be extremely high. The target spatial wavelength component is completely masked, which sometimes causes large measurement errors.

On the other hand, as a means to overcome such drawbacks, a method as shown in the above-mentioned Japanese Patent Laid-Open No. 154220/1983 has been proposed. That is, in this known example, the shape of the spatial filter detector is devised so that it has a linear taper in the direction of movement so that it can operate normally even in degrees ζ in response to signal inputs such as white lines.

However, this known example is a countermeasure for a very special input function and does not provide a general answer. As is clear from this assumption, an output similar to that of the known example shown in FIG. 12 will appear in this case.

As will be discussed further below, spatial filters can be thought of as acyclic digital filters that use a finite number of sample values, and therefore involve problems specific to digital filters due to handling a finite number of samples. The conventional example does not take this point into account.

An object of the present invention is to provide a spatial filter that improves the characteristics of the spatial filter, particularly its homomorphic characteristics, and enables highly accurate speed detection.

Means for Solving Problem C] The above object is achieved by applying a predetermined window function when extracting signals from each photodetecting element constituting the spatial filter.

[For production]

Spatial filter detector is a kind of acyclic type (FII?)
Since it works as a filter, frequency characteristics can be improved by applying a window function. For theoretical explanations, see, for example, A. Otzpenheim et al., 1 Digital Signal Processing, Prentice Hall (1975), 2
Pages 37 to 270 (A, V, Oppenheis
+ and R,W.

5chafer; “D 1g1tal S ign
al Processing Prentice-1
(all 1975. PP237-270) for details.

〔Example〕

Hereinafter, the spatial filter according to the present invention will be explained in detail with reference to illustrated embodiments.

First, FIG. 2(a) shows an example of the configuration of an optical system of a spatial filter for speed detection. A real image of 9 (for example, a road surface) is focused by lens 8 . However, the shape of the Pi surface is a detector in the form of brightness changes.
Each photodetecting element (hereinafter referred to as "element") 1
(1) to 1 (N) will be sampled.

Therefore, if these elements 1 (1) ~ are arranged at a constant pitch P as shown in Figure 2 (b), there is a constant distance between each of these elements adjacent to each other. There will be a delay element. That is, in a time axis filter, the delay is time, but in a spatial filter, it is distance. Note that these correspondence relationships become clear when considering the case of moving at a constant speed.For example, the road surface 9 in FIG.
Focusing on one point S, the image on the surface of the detector 1 becomes S'. Therefore, as shown in Figure (b), the image S' of the two points S is
(N) one after another at regular time intervals, and a signal based on this image S' is obtained one by one from each terminal X t '' X N.

Therefore, N-1 delay elements D as shown in FIG.
It can be treated as a model with ID, .

Here, Z-1 indicates a delay due to Z transformation.

Using this model, the transfer function of the conventional spatial filter shown in FIG. 12 can be obtained as follows. That is, in this FIG. 12, each element 1 (1) to
1 (N) are added with alternating positive and negative weights of 1, so the transfer function is H(Z)-1-Z-'+Z-''-Z-'...
・・・・・・・・・(1) N-I Z-18-1)・・
It is expressed as (1).

Therefore, the frequency characteristic is Z==6j in equation (1) expressing this transfer function.
(2) Here, j......Imaginary unit μ...
・・・・・・Spatial frequency (rad/m) P・・・・・・
...obtained by setting the element circumference as '#A(m). The results are shown in FIG.

As is well known, the filter having the transfer function shown in equation (1) is one of the acyclic filters (FIR) among digital filters.
), and it is known that strong side lobes remain in the frequency characteristics of the transfer function used to obtain output from a finite number of sample results.

On the other hand, in such an FIR filter, frequency characteristics can be improved by performing a type of weighting called a window function. In other words, the frequency characteristics of the spatial filter can be improved by treating it as a FIR type digital filter.

The present invention has been constructed from this point of view, and FIG. 1 shows one embodiment thereof. First, in FIG. 1(a), 1 is a light detector; Similarly, the elements 1 (1) to 1 (8) have the same dimensions and are arranged in a straight line at equal pitches. Note that the number of elements here is 8 for convenience of explanation, and it goes without saying that any number can be selected depending on the purpose.
) can be a photodiode, solar cell, or other photodetecting element (with sufficient sensitivity in the wavelength band used,
Any material may be used as long as the dimensions can be realized with high precision and each material has well-matched characteristics.

The outputs of each element 1 (1) to 1 (8) are respectively input to the weighting circuits WGI-WG8, and after being weighted by weighting coefficients g1 to g8, they are added by the addition circuit 20 and output. . here,
For example, as shown in FIG. 1(b), the weighting coefficients g1 to g8 are weighting functions W (X) and -W (
All you have to do is choose a clay sample of X). Note that the weighting function W (X) is known to have names such as Hamming, Hanning, Bartlett, Gauss, and Blackman, and may be selected depending on the purpose. Since the spatial filter is of a band-pass type, these weighting coefficients are selected so that they are alternately positive and negative, as shown in FIG. 1(b). Note that the number of elements here can be either even or odd, but if the number is odd, the positive and negative outputs will not be equal, so some DC component will appear, but it is easy to deal with it, and it is especially important to avoid problems. There isn't.

As an example, if a panning window well known as a weighting function W (X) is employed, a characteristic as shown in FIG. 4 is obtained. As is clear from comparing this characteristic with the characteristic of the conventional example shown in FIG. It can be seen that the pressure characteristics are improved by about twice as much.

Therefore, this embodiment has a great effect on the I wave of the road surface signal whose level increases as the frequency range decreases.

Note that although the width B of the main lobe is somewhat wider than the width B in FIG. 13, noise components near the center frequency do not pose much of a problem. Furthermore, by increasing the number of elements N, the main lobe width can be made as narrow as necessary, so this does not pose much of a problem.

Next, an embodiment of the weighting circuits WGI to WG8 will be described with reference to FIG. Note that this figure shows one (i-th) of the weighting circuits, and also shows an example in which a photodiode is used as a detector.

□ Since the output of photodiode 1 (i) is obtained in the form of a current, current-voltage conversion is performed using an operational amplifier AMP and a resistor Rfl.

Output y5 is 3't = RtAIゑ...
...(3), so the weight is given by -R25. At this time, since the weight value is relative, one of the resistances can be arbitrarily determined, and the others can be determined by the weight ratio. Note that as the number of elements N increases, the constant value also changes accordingly. Since a wide variety of values are required, it is advantageous to use thick film resistor printing techniques.

Next, the adder circuit 20 can be realized by a circuit configuration as shown in FIG. At this time, since it is an adder circuit, the resistors 20(1) to 20(N) are all set to the same value. Then, depending on the sign of the weight, the inputs may be connected to the positive input side or the negative input side of the operational amplifier 200.On the other hand, the resistors 210 and 211 may have the same resistance value. Dynamic range must be taken into account. Note that although this example uses a current output type element such as a photodiode for the detector L, if a voltage output element is used, the fifth
It goes without saying that the current-voltage conversion circuit shown in the figure becomes unnecessary, and the weighting in this case is determined by the resistors 20(1) to 20(N) in FIG. The output of the operational amplifier 200 for 5 is here, Rz+o + Re5t is the resistance value of the resistor with the corresponding symbol in FIG.
z. , becomes.

By the way, in the embodiment shown in FIG. 1, weighting is performed by an external circuit, so the detector 1 may be, for example, a device having elements of equal size commercially available as a photodiode array, or a graphic reader such as a facsimile machine. A line sensor using a charge-coupled device (CCD) or the like can be used.

Next, another embodiment of the present invention will be described using FIG. 7.

In the embodiment shown in FIG. 1, weighting was performed by an analog circuit, but in this embodiment, weighting is performed digitally, and each element l (1) to 1 (8
) are sequentially selected one system at a time by an analog multiplexer 502, and the outputs of the analog/digital converters (ADC
) 501. The ADC 501 converts the digital value into a digital value, which is then read by a microprocessing unit (MPU) 500, subjected to arithmetic processing, and output. Note that if an analog output format is required, it may be output via a digital-to-analog converter (DAC).

Next, an operation flowchart of the MPU 500 is shown in FIG.

First, in the first step 8, the variable S indicating the total value
Clear UM.

In step B, a loop counter (2) is set to an initial value of 1.

In step C, the signal of the detector element indicated by the loop counter I is sent to the ADC through the multiplexer 502.
501.

In step D, the output value V of the ADC 501 is read. Note that if the conversion speed of the ADC 501 is not fast enough, it is necessary to provide an appropriate means between steps C and D to set a predetermined time interval.

In step E, the output value V is multiplied by the weight gi corresponding to its tap and added to the total value SUM. Note that the weight gl may be stored in the internal memory of the MPU 500.

In step F, the loop counter is incremented by one, and in step G, it is determined whether the gt final value (number of elements N) has been exceeded. If not, the process returns to step C (repeat). Output the result SUM at H.

By the way, when the target surface (road surface) is moving relative to the spatial filter, the output changes temporally in a roughly sinusoidal manner. This period corresponds to the relative speed, and therefore the series of processes shown in FIG. 8 must be sufficiently faster than this period.

As described above, in the embodiment shown in FIG. 7, processing is performed digitally, so problems peculiar to analog circuits such as drift and component precision can be avoided.

A further feature is that it is easy to adaptively change the weighting coefficients. That is, by simply changing the software, it is possible to change the window function in response to other conditions.

Another feature is that the functions of a post-processing circuit such as an advanced center frequency estimating circuit or a variable bandpass filter as shown in the conventional example of FIG. 12 can be realized by software processing of Mpusoo. As a result, except for a very small portion of the circuits, digitization is achieved, and it becomes possible to significantly downsize the circuits by converting them to LSi.

Next, another embodiment of the present invention will be described with reference to FIG.

First, FIG. 9(a) shows the configuration of a spatial filter, and the difference from the conventional example explained in FIG. 12 is that the shapes of the detector elements 1) to 1(N) are different. As noted on the left side of the figure, the width of each element 1 (1) is a value obtained by sampling a function g (x) that gradually decreases from the center in the front-rear direction. Same figure (
b) shows the detailed structure of the detector elements 1 (1) to 1 (N), and only two elements are shown here for the sake of explanation.

Weighting of the sample values of the spatial signal can be given by the light receiving sensitivity of each element. Therefore, in this embodiment, the light-receiving sensitivity is changed by changing the light-receiving area of each element, so that weighting can be obtained.At this time, as is clear from FIG. The dimension (length) L in the direction is kept constant for each element, and the arrangement pitch P of the elements is also kept constant in order to extract a constant spatial wavelength. Dimensions (width) W
This can be obtained by changing +.

At this time, if the width W5 of this element increases, the direct current sensitivity increases proportionally, but at the same time a change appears in the sympathetic characteristics.

FIG. 10 shows the spatial frequency characteristics of a single element, and it can be seen that the characteristics change depending on the radiance of the element. As is clear from this figure, as the element's radius W5 increases, the high frequency component is attenuated, and the output becomes zero when the width W1 is an integral multiple of the wavelength.

In addition, as a matter of course, the width W! must be smaller than the pitch P, and the center frequency of the spatial filter is 1/2
Since it is given by P, the center frequency will be on the left side of the center of FIG. That is, the attenuation is only about 3 dB at most, but it cannot be ignored because the greater the number of elements, the higher the weighting accuracy is required.

The function g(x)- shown on the left side of FIG. 9(a) is applied in consideration of the above, and therefore, according to this embodiment, it is possible to provide a sufficiently appropriate window function. Since it is possible to eliminate the need for a weighting device outside the detector, the structure is simple and miniaturization can be achieved easily.

Next, another embodiment of the present invention will be described with reference to FIG.

The embodiment shown in FIG. 11 uses an optical filter for weighting, and as shown in FIG. 11(a), a filter 150 is provided on the light receiving surface of the detector 1. Note that this filter 150 is not intended for wavelength distribution (it is for changing the amount of light transmission, and is a type of so-called ND filter).

In FIG. 11(a), the detector 1 is the same as the conventional example shown in FIG. ing. As described above, the filter 150 whose light transmittance is distributed as shown in FIG. 2B is arranged above the detector l.

Therefore, the sensitivity of each element changes depending on the transmittance of the filter 150, and the transmittance function of the filter 150 becomes a window function.

Therefore, the difference between this embodiment and the conventional example is that the filter 150
Therefore, the performance of the conventional spatial filter device can be significantly improved simply by adding the filter 150.

In this example, the transmission characteristic of the filter 150 is a continuous function as shown in FIG. It may also be used as an example.

[Effects of the Invention] According to the present invention, it is possible to significantly improve the frequency characteristics of a spatial filter simply by using digital filter theory, which not only improves measurement accuracy but also improves post-processing for center frequency estimation. This has the effect of simplifying the circuit.

To explain in more detail, in the conventional example shown in FIG. 12, even if the number of elements N is significantly increased, the spatial frequency characteristics are not improved to that extent. In this case, the width B of the main lobe is Although it decreases as N increases, the suppression ratio of the first side lobe remains almost unchanged at about 13 dB, and the suppression ratio of other side lobes is not expected to improve significantly. For example, even if the number of elements N is 100, , the suppression ratio near DC is as small as about 40 decibels.

On the other hand, according to the present invention, even in the case of the embodiment using the Hanning function as the window function, the number of elements is the same, N-100
In this case, a suppression ratio of 100 decibels or more in the vicinity of DC can be obtained, and a large performance improvement can be obtained.

[Brief explanation of the drawing]

FIGS. 1(a) and (b) are a configuration diagram and characteristic diagram showing one embodiment of a spatial filter according to the present invention, and FIGS. 2(a) and (b) are
1 is a configuration diagram of the optical system of a spatial filter and an arrangement diagram of elements, FIG. 3 is a model diagram of a spatial filter, FIG. 4 is a characteristic diagram of an embodiment of the present invention, and FIG. 5 is an embodiment of a weighting circuit. 6 is a circuit diagram showing one embodiment of the adding circuit, FIG. 7 is a block diagram showing another embodiment of the present invention,
FIG. 8 is a flowchart for explaining the operation, FIG. 9(a),
(bl is an explanatory diagram showing yet another embodiment of the present invention,
Figure 10 is a curve diagram showing the frequency characteristics of the element.
．． 1(a) and 1(b) are a block diagram and characteristic diagram showing still another embodiment of the present invention, FIG. 12 is a block diagram showing a conventional example of a spatial filter, and FIG. 13 is a characteristic diagram of the conventional example. It is. 1......Optical detector, 1 (1)~l
(N)...Element, 20...
... Addition circuit, WGI to WGQ ...
・Weighting circuit. Fig. 1 Fig. 2 Fig. 3 XL U Rectangular shape Figure 12 Figure 13 Space circumference liquid number (Kanshuang old)

Claims (1)

[Claims] 1. A photodetecting element group consisting of a plurality of photodetecting elements arranged in a straight line at regular intervals, converting an image of the surface of an object moving relative to itself into an electrical signal, In a spatial filter that detects relative velocity based on frequency components included in a signal, a weighting means is provided for setting an independent weighting coefficient for the photodetection sensitivity of each of the plurality of photodetection elements, and 1. A spatial filter characterized in that it is configured to provide a light detection sensitivity that gradually decreases from the center toward both ends in the arrangement direction of the elements. 2. The spatial filter according to claim 1, wherein the weighting means is constituted by an electric circuit provided on the output signal side of the photodetecting element. 3. The spatial filter according to claim 1, wherein the weighting means is constituted by an optical filter arranged on the photodetection surface of the photodetection element group. 4. The spatial filter according to claim 1, wherein the weighting means comprises means for changing the light-receiving area of each of the plurality of photodetecting elements. 5. The spatial filter according to claim 2, wherein the electric circuit is constituted by a counting processing circuit. 6. The spatial filter according to claim 4, wherein the change in the light-receiving area is given by a dimensional change in the arrangement direction of the plurality of photodetecting elements. 7. In claim 2, the electric circuit comprises:
A spatial filter characterized by being composed of an operational amplifier and a thick film resistor element.