JPH0711527B2 - Speed measuring device - Google Patents

Description

Description: TECHNICAL FIELD The present invention relates to a speed measuring device using an image sensor, and in particular, accurately captures a frequency of a signal subjected to spatial processing, thereby achieving high accuracy in speed measurement. The present invention relates to a speed measuring device suitable for performing.

[Conventional technology]

The conventional device, as described in JP-A-5-14044,
In velocity measurement using a spatial filter, threshold levels are provided at two locations, positive voltage and negative voltage, in order to eliminate the small amplitude portion of the spatially processed narrowband irregular signal (which causes frequency disturbance). The signal was ternary and the frequency was measured. However, no distinction is made between non-moving or stationary and moving.

[Problems to be solved by the invention]

As described above, the related art describes the signal processing of ternarizing a narrowband irregular signal, and does not consider the timing of signal acquisition.

An object of the present invention is to eliminate the timing at which a signal is taken in and a portion where the amplitude level of a narrowband irregular signal is small, and to measure with high accuracy not only in the constant velocity region of a moving object but also in the acceleration / deceleration region at start and stop. Is what you are trying to do.

[Means for solving problems]

The above-mentioned object is a line sensor for inputting reflected light from an object to be measured through a lens, and an effective pixel of the line sensor is S
When 1, S2, S3, S4, ..., Sn (n = 1,2, ...), the first system is S1, S5, S9, ..., S4n + 1 (n = 0,1,2, ...) , The second system is S3, S7, S11, ..., S4n + 3 (n = 0, 1, 2, ...) And the video signals of the effective pixels of the first and second systems are adopted, S
2, S4, S6, ..., S2n (n = 1,2, ...) Spatial filters configured not to adopt video signals of effective pixels, and A / D for A / D converting the video signals of each pixel. A D converter, an adder for obtaining the total sum of the first system and the second system, a subtracter for obtaining a difference between the total sum of the first system and the second system, and an output signal of the subtractor D / A / D converter for A conversion, an integrator for integrating the output signal of the D / A converter, a detector for full-wave or half-wave rectification of the output signal of the integrator, and the detected signal and threshold And a means for measuring a frequency only when the detected signal is equal to or higher than a threshold voltage, and a means for obtaining a center frequency by frequency distribution processing of the measured frequency data. Achieved by

[Action]

The narrow-band random signal of the spatial filter has no center frequency and a noise level when the moving object is stationary. Also, when speed occurs, the signal level immediately rises. Therefore, as described above, it is possible to determine whether speed has occurred by comparing the narrow band irregular signal with the threshold voltage that has been rectified and detected, and it is also possible to remove the narrow band irregular signal that is equal to or lower than the threshold voltage. .

Therefore, in such a device, even when the moving object suddenly stops or slips, a signal corresponding to the speed can be immediately obtained.

Moreover, the frequency measurement of the narrow band irregular signal is performed by shaping the narrow band irregular signal into a rectangular wave pulse through the zero cross detection circuit,
The method is such that the cycle is counted by another stable high frequency pulse, and is obtained from the count value. If the count value is not fetched when it is determined by the signal discriminating apparatus that the signal is not a signal, a stopped state or a portion of the narrow band irregular signal having a low level is removed. Further, the count value, that is, the frequency data, is subjected to frequency distribution processing, etc., and the accuracy of speed measurement for obtaining the center frequency is improved.

〔Example〕

 An embodiment of the present invention will be described below with reference to the drawings.

FIG. 1 is a block diagram of a speed measuring device of the present invention.
An image sensor 1 is, for example, a CCD line sensor. The line sensor has (1) a small pixel pitch,
(2) Each dimension of the light receiving part can be made with high accuracy. (3) The signal of each pixel can be sequentially read by the clock. There are advantages such as. Reference numeral 2 is an amplifier, and in the case of a CCD line sensor, it is a differential amplifier that extracts a light reception signal by differentially outputting two output signals of the sensor. Reference numeral 3 is an A / D converter for converting the light receiving level of each pixel of the line sensor 1 into a digital value. Reference numerals 4 and 5 denote buffers with a gate, and the gate counts only the number of dummy pixels (6, 8) from the reference clock φ15 for reading the light receiving level of each pixel of the line sensor and the scan signal SH17,
It is turned on and off by clocks TA7 and TB9 having a pitch cycle of the spatial filter. This signal will be described later. Numerals 10 and 11 are adders for adding the digital values of the pixels passed by the clocks TA and TB. When the total sum of two series of one scan (A and B series for convenience) is obtained, subtraction 12 is carried out by the scan clock. The output of the subtracter 12 is pitch-shifted, that is, amplified 13, and a signal obtained by delaying the scan clock 18 is converted into an analog signal by the D / A converter 14. The output of the D / A converter is delayed and integrated 19 with the value of each cycle of the scan clock 17 to form a continuous waveform. At this time D /
When the latch time of the A converter approaches the scan time, the integrated signal level rises. This integrated signal presents a narrow band random signal.

Here, there is also a method of reading the subtracted digital value for each scan clock into a computer and obtaining the center frequency using software such as FFT (Fast Fourier Transform).

High frequency reference clock 15 for narrowband irregular signals
Since signals such as and scan clock 17 are superimposed,
High frequency components are removed via LPF (Low Pass Filter) 20. Although not shown in the figure, the DC component is also removed by using a capacitor. The low frequency range is passed to measure the frequency of the low speed range.

The signal from which noise has been removed branches into a frequency measuring device and a signal discriminating device. In the frequency measurement, the narrow band irregular signal is shaped into a rectangular wave pulse by the zero-cross detection and the comparator circuit 21, and the period of this signal is counted by using the counter 23 with another stable high starting frequency 22 pulse. The counted value is connected to the microcomputer 31 via the latch circuit 24.

Further, in the signal discrimination, the signal after the LPF 20 is full-wave 26 or half-wave rectified and detected 27 through the buffer 25 to obtain an envelope component signal. This detected signal is the threshold voltage
The Vref is compared with the comparator 28. The level of the threshold voltage Vref is experimentally set to a level higher than the noise level when the moving object is stopped and slightly higher than the voltage level at which the frequency wave of the narrowband irregular signal is disturbed. The output signal of the FF (flip-flop) circuit 29 which catches the rising edge of the output signal of the comparator 28 and the pulse waveform signal of the irregular signal is input to the AND circuit 30. The output signal of the AND circuit 30 is input to the microcomputer 31.

The microcomputer 31 reads the counter data by looking at the output of the AND circuit 30. That is, when the output is “H”, the data is read, and the data latch circuit 24 and FF circuit
Clear 29. If the data is not read when it is “L”, the data is not read when the moving object is stopped and when the level of the irregular signal is small, and the reliability of the data is improved.

Next, the principle of the non-contact speedometer using a spatial filter will be briefly described. FIG. 2 schematically shows the relative relationship with the lens 33 when the light receiving elements 32 arranged in a line are moving at the speed V with respect to the subject 34. The principle is that random reflection unevenness on the surface of the subject 34 is optically detected, spatial processing is performed, and the velocity V of the moving object can be measured in the form of frequency. Here, the spatial processing means that the photoelectric conversion elements 32 are arranged in a line at regular intervals as shown in FIG. 2 and the sum of photocurrents I _A and I _B connected every other line.
Is to take the difference. This signal exhibits a narrow band random signal. The frequency of this narrowband random signal When the velocity of the moving object is V, the magnification of the lens is m, and the pitch of the light receiving elements 32 (pitch of the spatial filter) is P, the following relational expression is obtained.

That is, since the pitch P and the magnification m are known, the velocity V can be measured by obtaining the center frequency c.

Next, the structure of the spatial filter when the CCD line sensor 35 shown in FIG. 3 is used will be described.

The line sensor 35 has a pixel pitch of several tens of μ as described above.
m is small, and the gap between pixels is very narrow, such as several μm. Therefore, when an object is observed by using an optical system, the resolution is good, so that the uneven pattern of the object can be captured with high precision, and the moving speed of the object with small unevenness can be measured.

The spatial filter is constructed by taking out pixel signals arranged in a line and skipping every one or every few pixel signals by a pulse train as shown in FIG. Third
The figure shows the case where the filter pitch P is 4 pixels.

Next, a method of forming a spatial filter with a pulse train will be described with reference to FIG.

In FIG. 4, (a) SH signal 36 is a scan clock signal, and (b) φ signal 37 is a clock for reading the pixel signal of the line sensor, which is the reference clock of FIG. (A)
The SH signal is created by counting 16 pixels of the line sensor. The signal (c) 38 is an output signal of the line sensor and is a signal whose level varies depending on the amount of light received by the subject. In the case of a CCD line sensor, generally, the signal of (c) 38 and the signal of (c) 38, which is not shown in FIG. 4, are synchronized and actuated with a pulse signal of a certain level to eliminate the offset noise. Then, when amplified and integrated, the uneven pattern of the line of the subject shown in (d) 39 can be observed.

Generally, CCD line sensors and the like have dummy pixels, that is, insensitive pixels, at both ends.

(E) The TA signal 40 is a pulse signal that counts 6 dummy pixels from the scan clock SH and is every 4 pulses (1 pulse corresponds to 1 pixel) from the beginning of the signal pixel. () TB
Similarly, the signal 41 is a pulse signal that counts (dummy pixels + 2 pixels) 8 from the scan clock SH and is every 4 pulses. (G) 42 represents the pixels of the line sensor, and the number of signal pixels is 128.

If the signal pixels are S ₁ , S ₂ , S ₃ , ..., S ₁₂₈ , the pixels corresponding to the TA signal are S ₁ , S ₂ , ..., S ₁₂₅ , and the pixels corresponding to the TB signal are S ₃ , S ₇ ..., S ₁₂₇ . Therefore, in Fig. 1, TA, TB
When the digital value of the line sensor after A / D conversion is passed using the pulse of, the signals of the pixels S ₂ , S ₄ , S ₆ , ..., S ₁₂₈ are not captured and a filter is formed.

In FIG. 4, the pitch P of the spatial filter is 4 pixels, but since the pulse trains of (e) and () can be arbitrarily selected, the pitch P can be easily changed.

Further, in FIG. 1, for all pixels of the line sensor
A / D converted, but A / D at timing of TA and TB signals
There is no problem in converting and adding.

The above is the configuration of the spatial filter using the line sensor.

Next, the narrow band irregular signal will be described.

FIG. 5 (h) 43 shows the waveform of the narrow band irregular signal after LPF when the moving object is moving at a constant speed. An irregular signal has a low vibration level and a disturbed frequency. Further, the line sensor is installed so that the line direction of the line sensor coincides with the moving direction and is parallel to the object surface. If this is not the case, the irregular signal becomes unsymmetrical, that is, a waveform in which low-frequency components are superimposed, and the frequency cannot be measured accurately. A method of measuring the frequency of a narrow band irregular signal will be described with reference to FIG. 6 using a pulse waveform. (I) 44 is an irregular signal, and (j) 45 is a pulse signal shaped into a rectangular wave by zero crossing. (K) 46 is an oscillation frequency pulse for counting, and a crystal oscillator or the like is used to stabilize the oscillation frequency.
As an example of the method of counting the period of the pulse waveform of (j) 45 with the signal of (k) 46, a method of counting using two series of counters will be described.

First, the first series counter counts from the rising of the (j) 45 signal to the next rising ((1) 47 signal). Next, from the rising edge to the rising edge, the second series counter counts ((m) 48 signal). That is, (j) 45 in 2 series
In this method, it is possible to capture all the cycles. The frequency is represented by count value / oscillation frequency.

Next, using FIG. 7, it is determined whether or not the signal is based on speed,
We describe a system that eliminates the low level of narrowband random signals.

The (h) 49 signal is a narrow band irregular signal, and the (o) 50 signal is a signal obtained by full-wave rectifying the (h) signal. The (p) 51 signal detects the envelope with the signal waveform obtained by detecting the (o) signal. The level relationship between the (p) signal and the threshold voltage Vref is also shown. And when the moving object is stopped, the noise level is almost zero because the DC is cut. Therefore, as in (p) 51, the threshold voltage Vref is set to a level slightly higher than that where the amplitude level of the narrowband irregular signal is small. The (p) signal and Vref are input to the comparator, and the output waveform is (g) 52. Therefore, the irregular signal (h) 49 is pulse-shaped at zero cross, and the period read by the computer 31 is the signal shown in (r) 53 in the state where the (q) signal rises. By doing so, it is possible to remove the frequency where the amplitude level is small where the frequency is disturbed by the narrow band irregular signal. Also, it is possible to determine the presence or absence of an irregular signal.

A case where the system shown in FIG. 1 is mounted on, for example, a self-propelled robot will be examined.

(1) When the self-propelled robot goes straight, it takes a short time until the command "go straight" is output and the driving motor for running actually starts moving. Therefore, as described above, in the pulse count frequency measuring method, the frequency is measured from the noise signal within the above time.

(2) When the self-propelled robot stops, it takes a certain time until the command "stop" is output and the robot completely stops. Therefore, unless the spatial frequency data is read at the same time as the "stop" command, an error will occur.

(3) When the self-propelled robot is traveling straight ahead, if all vehicles slip and the robot itself does not move in an extreme case,
The output of the spatial filter is the noise level, and the frequency is measured in this state.

As described above, in the cases of (1) to (3), reliable data can be obtained by observing the level of the output (irregular signal) of the spatial filter and capturing the spatial frequency.

The data of the spatial frequency when moving by the above method will be described below.

A reliable spatial frequency can be obtained using the system shown in FIG. 1, but there are variations as shown in FIG. 8 when the spatial frequency is experimentally measured. Therefore 1
By calculating the velocity from the individual spatial frequency data,
As shown in the figure, the reliability of the data is improved when the frequency distribution is taken and the peak frequency thereof is set as the center frequency c.

Further, the moving speed is experimentally changed, the center frequency c is obtained from each frequency distribution when the moving speed is a certain fixed distance, and the speed is calculated from the equation (1). From the accuracy of this speed and the actual average speed, the accuracy is slightly improved when the center frequency is the average value of c ± Δc.

Therefore, speed measurement of moving objects such as self-propelled robots is the 10th
As shown in the figure, the central frequency c is first obtained by the frequency distribution, and from then on, the average value within c ± Δc is newly set as the central frequency c.

With such an algorithm, it is possible to follow the change in speed and measure the speed with high accuracy.

〔The invention's effect〕

As described above, according to the present invention, it is possible to measure the spatial frequency only when the object is moving, and it is possible to eliminate the case where the level of the narrowband irregular signal is small, and improve the signal property of the spatial frequency on the circuit. In addition, the center frequency was obtained from the frequency distribution processing of the spatial frequency data in software. As a result, highly accurate speed measurement was possible.

[Brief description of drawings]

FIG. 1 is a system block of an embodiment of the present invention, FIG. 2 is a conceptual diagram of an optical system using a spatial filter, FIG. 3 is a configuration diagram of a line sensor and a spatial filter, and FIG. Illustration of input / output signals and pulse signals of filter configuration, Fig. 5 is a narrowband irregular signal waveform diagram, Fig. 6 is a frequency measurement signal waveform diagram, and Fig. 7 is a small signal level portion of Fig. 5. Fig. 8 is a signal waveform diagram for eliminating the noise, Fig. 8 is an explanatory diagram of spatial frequency data obtained in the experiment, Fig. 9 is its frequency distribution diagram, and Fig. 10 is a flow chart of speed measurement in a moving vehicle. 1 ... Image line sensor, 3 ... A / D converter, 10, 11 ...
… Adder, 12 …… Subtractor, 14 …… D / A converter, 23 …… Frequency measurement counter, 26 …… Full wave rectification, 27 …… Detection, 28 …… Comparator, 31 …… Microcomputer , 43 …… Narrowband irregular signal.

Claims (1)

[Claims] 1. A line sensor for inputting reflected light from an object to be measured through a lens, and effective pixels of the line sensor are S1, S2, S3, S4, ..., Sn (n
= 1,2, ...), the first system is S1, S5, S9, ..., S4n
+1 (n = 0,1,2, ...), 2nd system S3, S7, S11, ..., S4n
+3 (n = 0, 1, 2, ...) And the video signals of the effective pixels of the first and second systems are adopted, and S2, S4, S6, ..., S2n (n =
1, 2, ...) Spatial filter configured not to adopt the video signal of the effective pixel, an A / D converter for A / D converting the video signal of each pixel, the first system and the second system An adder for obtaining the sum of the above, a subtracter for obtaining the difference between the sum for the first system and the sum for the second system, a D / A converter for D / A converting the output signal of the subtractor, An integrator for integrating the output signal of the A converter, a detector for full-wave or half-wave rectification of the output signal of the integrator, a comparator for comparing the detected signal with a threshold voltage, and the detected A velocity measuring device comprising: a measuring unit that measures a frequency only when a signal is equal to or higher than a threshold voltage; and a unit that obtains a center frequency by frequency distribution processing of the measured frequency data.