JP3362707B2 - Temperature change and drift coefficient estimation device for thermal infrared imaging device - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an infrared image pickup device driving device, and more particularly to a thermal type infrared image pickup device driving device which captures incident infrared rays in the form of heat. The present invention relates to correction of so-called drift that changes with time.

[0002]

2. Description of the Related Art The present invention is an improvement of the invention described in Japanese Patent Application No. 7-75264, which is filed by the same applicant as the present application which is the prior invention of the present inventor.

The thermal infrared imaging device of the above-mentioned prior application has a semiconductor substrate 700 and a scanning circuit 701 on the surface of the substrate, as shown in FIG. 7, and a light receiving portion for converting incident infrared rays into an electric signal thereon. have. This circuit and the light receiving unit integrate a plurality of pixels, that is, thermoelectric conversion elements, so that a two-dimensional infrared image can be obtained. The light receiving portion is composed of an infrared absorption layer that absorbs infrared rays, a diaphragm that prevents heat from escaping, and a thermoelectric conversion element that converts heat into an electric signal. The diaphragm forms a film-like structure floating in the air by removing the underlying layer by etching. In this example, the thermoelectric conversion element uses a bolometer 707 whose electric resistance value changes with temperature, and titanium is used as the bolometer. Infrared rays that have entered each pixel are absorbed by the infrared absorption layer of each pixel and raise the temperature of the diaphragm of each pixel. This temperature rise is converted into an electric signal by the titanium bolometer 707 and sequentially read out through the circuit on the substrate.

In general, in an infrared image pickup device, regardless of whether it is a thermal type or a quantum type (photoelectric effect type using a semiconductor), there are variations in bias level due to variations in the detector of each pixel. This is fixed pattern noise (FPN)
And is normally corrected by a correction circuit. As an example of this, as described in the above-mentioned prior invention, a memory for holding the variation amount of the bias level is provided to correct the fixed pattern noise.

Further, in the quantum thermal infrared imaging device, as shown in FIG. 9, there is a so-called sensitivity variation in which the relationship between the infrared input energy and the detector output varies depending on the pixel. In such a state, if the background temperature of the subject changes and the infrared input energy of the entire screen changes,
Even if fixed pattern noise is removed at x0 in FIG. 9,
The fixed pattern noise corresponding to Δy1 appears again due to the sensitivity variation. As a countermeasure against this, a gain correction circuit configuration as shown in FIG. 8 is provided, and the sensitivity variation Δy2 / y2 is held in advance in the sensitivity difference memory.
.DELTA.y1 = y1.multidot..DELTA.y2 / y2 is calculated for the detector output y1 and added to y1 to correct the fixed pattern noise due to such sensitivity variation (for example, Japanese Patent Laid-Open No. 2-107074). reference).

Further, in the thermal infrared detector, for example, as shown in Japanese Patent Laid-Open No. 7-193752, an example is proposed in which a pixel so-called optical black (OB) that does not feel incident infrared light is provided to make it a reference level of a signal. ing.

[0007]

However, the above-mentioned prior art has the following drawbacks.

A thermal infrared detector using a bolometer as shown in FIG. 7 has a problem peculiar to the bolometer type in that when the temperature of the device changes, the temperature change remains a signal. Moreover, since the amount of change varies depending on the sensitivity variation of each pixel, even if fixed pattern noise (hereinafter abbreviated as FPN) is removed at a certain point, the FPN reappears due to temperature change. The sensitivity variations are not small because they are caused by variations in the temperature coefficient of the bolometer, variations in the resistance value itself, and the like.

For example, since the temperature change of the diaphragm with respect to the temperature change of the object is minute, there is a signal change corresponding to the change of the object by about 500 ° C. due to the change of the device 1 ° C. If the sensitivity variation of each pixel is set to 5%, the FPN will be about 25 ° C. in terms of the object. When no temperature control is performed, the device temperature changes by several degrees C. immediately after the power is turned on and by about 1 degree C. even after being stabilized.

On the other hand, in the case of applying the FPN correction in consideration of the sensitivity variation, which is performed in the conventional quantum infrared detector, there are the following problems.

The thermal infrared detector generally needs to amplify a large signal because of its low sensitivity. However, since the drift of the amplifier that does this is not negligible with respect to the signal level, the absolute level of the signal is measured. I can't. That is, only the relative level between pixels can be known.

Further, a reference pixel which is not sensitive to incident infrared rays which should be a reference level of a signal, that is, an optical black (hereinafter abbreviated as OB) is also affected by the temperature change of the device similarly to the effective pixel. Will end up.

In other words, the thermal infrared detector does not have a reference level for data processing, and the extension of the prior art cannot correct the FPN caused by the sensitivity variation.

In the infrared detector of the infrared image pickup device, variations in OB pixels and effective pixels occur due to changes in temperature, and the amount of this variation reflects changes in temperature of the device.

The present invention has been made in view of the above-mentioned conventional circumstances. Therefore, the object of the present invention is to provide an O detector for an infrared detector.
To provide a new thermal infrared imaging device capable of eliminating the above-mentioned drawbacks inherent in the conventional technique by correcting the variation due to the temperature of the effective pixel by using the variation due to the temperature of the B pixel. It is in.

[0016]

In order to achieve the above object, a thermal infrared imaging device according to the present invention comprises a plurality of reference pixels, which are insensitive to incident infrared rays, so-called optical black, and a plurality of effective pixels. It is characterized in that it has a memory for storing a coefficient of drift of each pixel and a drift correction means, and corrects the drift amount of the effective pixel based on the variation amount of the plurality of reference pixels.

The present invention also has means for obtaining the average of the output levels of the plurality of reference pixels and means for obtaining the displacement of each pixel from the average.

The present invention also has means for obtaining an average of absolute values of displacements of the reference pixels.

Further, the present invention has a memory for storing an average of absolute values of displacements of the respective reference pixels obtained at the time of initial setting.

Further, the present invention also has a memory for storing the displacement of each effective pixel obtained at the time of initial setting.

Furthermore, the present invention further comprises means for dividing the average absolute value of the displacement of each reference pixel which comes in real time by the average absolute value of the displacement of each reference pixel obtained at the time of initialization.

Further, the present invention has means for multiplying the division result by the displacement of each effective pixel obtained at the time of initial setting, and means for subtracting each obtained multiplication result from each effective pixel.

In other words, the thermal infrared imaging device according to the present invention blocks the incident infrared rays at the time of initial setting and
After correction, the temperature of the device is changed to calculate the average value of the output levels of a plurality of reference pixels that are not sensitive to the incident infrared rays by the OB pixel average value calculation circuit, and all of the reference pixels and the plurality of effective pixels are included. The average value of the output level of the reference pixel is subtracted from the pixel signal by the first arithmetic unit to obtain the drift coefficient for all the pixels, and for the reference pixel, the average value of the drift coefficients of the reference pixels is calculated, and in the normal operation, An average value of the output levels of the reference pixels is calculated, a displacement from the average value of the reference pixel output levels of the reference pixel signals arriving in real time is obtained by the first arithmetic unit, and an average of the displacements of the reference pixels is calculated. Value OB
The average value of the displacement of the reference pixels is calculated by a second average calculator, and the average value of the displacement of the reference pixels is divided by the average value of the drift coefficient of the reference pixels obtained at the time of initial setting. A third arithmetic unit multiplies the effective pixel drift coefficient obtained for the effective pixels, and the fourth arithmetic unit subtracts the multiplication result from the effective pixel signal that arrives in real time.

[0024]

As described above, the drift is a phenomenon in which the DC level of each pixel fluctuates with time, and if the fluctuation amount is the same in all pixels, the drift is caused by using the level of a reference pixel such as OB as a reference. Can be removed. However, in the thermal infrared sensor using the bolometer, the OB cannot be simply used as a reference because the drift amount of each pixel varies depending on the sensitivity variation between pixels as described above.

Drift is due to ambient temperature variations, which are approximately linear with temperature. That is, although the variation amount varies depending on the pixel, if there is a variation amount Δvi in the pixel i due to a certain temperature change ΔT, there will be a double temperature variation 2ΔT and a double variation amount 2Δvi. That is, if the variation amount of each pixel in a certain temperature change, that is, the so-called drift coefficient and the temperature variation amount are known, it is possible to return to the original screen without variation by subtracting them.
The reference pixel, so-called OB, blocks the incident infrared rays.
Variations among multiple OBs reflect temperature changes,
The temperature change can be estimated by measuring the variation of OB.

[0026]

BEST MODE FOR CARRYING OUT THE INVENTION The present invention will now be specifically described with reference to the drawings for each of its preferred embodiments.

FIG. 1 is a block diagram showing a first embodiment of a thermal infrared ray image pickup device according to the present invention.

Referring to FIG. 1, an infrared detector 101
Is a part of the conventional titanium bolometer type infrared detector,
Reference pixel that is not sensitive to incident infrared light, that is, OB
It is provided with pixels. The OB pixel is realized here by providing a shield plate that blocks incident infrared rays on the package. As the OB pixel, the first pixel to be read is desirable as shown in FIG. It is sufficient to have one in each row, but several rows can be formed as a spare in case of a defect. The method of providing the shield plate on the package is advantageous in that no change is made to the device itself, and there is no difference in structure between the OB pixel and the effective pixel except that the incident infrared rays are shielded.
The influence of disturbances other than incident infrared rays can be minimized.

Reference numeral 102 denotes an analog FPN correction circuit. The FPN correction circuit 102 removes FPN to some extent so that the level of each pixel falls within the dynamic range of the A / D converter 104. 105
Is a digital FPN correction circuit, which removes the FPN that could not be exhausted by the analog FPN correction circuit 102 to make a screen in which variations can be ignored.

In the present invention, further, the drift correction circuit 11
Have 7. The drift correction circuit 117 is further made up of several blocks and is shown in FIGS.
The operation will be described together with the diagram showing the operation of FIG. 2A and 2B show changes in the detector output of each of the OB pixel group and the effective pixel group as a function of the device temperature Td. Immediately after the FPN correction, the detector outputs of both the OB pixel group and the effective pixel group are complete. When the temperature of the device changes, the output of the detector changes, and the output varies depending on the sensitivity variation of each pixel. Since the sensitivity of each pixel is linear, the variation amount of the signal also increases linearly.

Here, as an initial setting, the device temperature is changed with the incident infrared rays blocked. This change may be changed by a Peltier element or the like, or a natural change in device temperature after power is turned on may be used. The temperature to be changed does not matter, and it suffices that the effective number of bits for obtaining the drift coefficient varies. When there is some variation, the drift coefficients of all pixels including OB pixels and effective pixels are acquired. This is because the OB average value calculation circuit 110 calculates the average value [vOB] of the output levels of all the OB pixels, and the subtractor 111 subtracts [vOB] from each pixel including the OB pixel and the effective pixel. It is stored in the OB section of 116. This Δvi0 becomes the drift coefficient of the pixel i. The drift coefficient of the OB pixel is specifically called Δvi0, OB. The drift coefficient of the effective pixel is Δvi0. Drift coefficient Δv of effective pixel
i0 is also obtained in the same manner as the OB pixel, and is stored in the effective pixel portion of the drift coefficient memory 116. To obtain the drift coefficient, it is better to average the results of several frames in order to reduce the influence of random noise. OB
The partial drift coefficient average value register 113 stores a value [Δvi0, OB] obtained by averaging the absolute values of the drift coefficients Δvi0, OB of each OB pixel.

In other words, assuming that the number of OB pixels is NOB,

[Equation 1] To calculate the average value register 113 of the OB drift coefficient
Store in. For this calculation, as indicated by the broken line in FIG.
The B displacement average value calculation circuit 112 may be used, or an external arithmetic device or the like may be used.

The above is the initial setting with the incident light being cut off, and the normal operation is performed as follows. As before, the OB average value calculation circuit 110 calculates the average value [vOB] of the output levels of the OB pixels at that time, and the OB displacement average value calculation circuit 112 calculates the average value of the output level of the OB pixels [vOB].
The average value [Δvi, OB] of the absolute value of the displacement from [OB] is calculated. This value represents the variation amount of the OB pixel and reflects the device temperature. That is, the absolute value of the displacement of the OB pixel changes every moment according to the drift. OB
The average value [Δvi0, OB] of the absolute value of the drift coefficient Δvi0, OB of the OB pixel acquired by the initial setting is stored in the average value register 113 of the partial drift coefficient, and the divider 114
Xd = [Δvi, OB] / [Δvi0, OB] This value xd indicates at what level the device temperature is currently at the time of initial setting, and the drift coefficient of the effective pixel acquired in the initial setting is added to this value xd.
By multiplying by 5, the drift amount of the effective pixel can be estimated. Since the effective pixel normally receives the incident infrared ray, it is unavoidable to make such an estimation. By subtracting the estimated drift amount of each effective pixel from each effective pixel by the subtractor 107, the drift correction is completed.

Although the frequency of this correction depends on the level of drift and the time constant, it may be performed once every several frames. The scan converter 108 is used for matching the operation timing of the infrared detector 101 and the NTSC output timing. An NTSC composite signal is output by the D / A converter 109 and other circuits.

The drive circuit 118 applies a pulse such as a clock for moving the infrared detector 101. Reference numeral 119 is a Peltier element for keeping the device temperature of the infrared detector 101 constant, and 120 is its controller. Temperature control usually has a fluctuation of about 100 mK,
Moreover, the drift does not disappear because of fluctuations in amplifiers other than the device, but the drift is considerably reduced. This allows a margin in the design of the dynamic range of the drift correction circuit 117. For the same purpose, the whole or a part of the circuit shown in FIG. 1 can be put in a temperature-controlled unit such as a thermostat.

FIG. 3 is a block diagram more specifically showing the drift correction circuit 117 shown in FIG.

Referring to FIG. 3, the infrared detector 101 is assumed to be a two-dimensional sensor having a size of about 128 × 128, one occupying each row, and a total of 128 OB pixels. The signal resolution is assumed to be 14 bits. O
The B average value calculation circuit 110 includes an ADDER 301, a latch 302, a RAM 303, and a timing generator 3 for the OB portion.
It is composed of 04. OB part timing generator 30
128 OBs by controlling each circuit by 4
128 is obtained by calculating the total of
It is divided by, and the average value of OB is obtained. The value is stored in the RAM 303.

The output of the subtraction circuit 111 is the OB of each OB.
The deviation from the average value appears. This displacement, that is, the variation of OB is assumed to be about 1/16 with respect to the dynamic range of 14 bits, and the number of bits here is 10 b.
It is enough.

The OB displacement average value calculation circuit 112 further averages the absolute values of the 128 OB variation amounts. The absolute value may simply take the sign bit. Here, the addition result is not divided by the number of OBs in order to obtain a large significant digit in the subsequent division circuit 114. That is, the average value [Δvi, OB] of the OB variation amount is multiplied by 128, and further shifted by 3 bits, and multiplied by 1024 in total.

Average value register 1 of drift coefficient of OB section
In 13 is stored the average value [Δvi0, OB] of the absolute values of the OB variation amounts acquired at the time of initial setting as described above. The output [Δvi, OB] of the average value calculation circuit 112 of the OB displacement that comes in real time is divided by the average value [Δvi0, OB] acquired at the time of this initialization (114), and how much it drifts from the time of the initialization. Ask for something.
At this time, since the real-time data [Δvi, OB] is multiplied by 1024, if the drift amount is the same as that at the time of initial setting, the value here is 1024. Also, [Δvi, OB] should be as large as twice as large as [Δvi0, OB].
The output of the division circuit 114 is 11 bits.

The output of the division circuit 114 is multiplied by the drift coefficient Δvi0 of each effective pixel acquired at the initial setting (115), and the drift amount Δvi of each effective pixel can be estimated. The reason why the upper 11 bits of the output of the parallel multiplier 115 are taken is to discard the lower 10 bits and restore the original 1024 times multiplication. The address generator 309 is
Addresses of effective pixels are sequentially generated.

When constructing the thermal infrared imaging device, it is necessary to consider how many bits should be assigned to the minimum temperature resolution. If it is over-allocated, the scale of the device will increase. Further, the FPN needs to be sufficiently small with respect to random noise in order to maximize the performance of the device. FPN is an effective value of 1 against random noise
If it is about / 4,

[Equation 2] It only needs to increase the total noise. FPN control is 1/1 of random noise (corresponding to the minimum temperature resolution)
When performing with an accuracy of 4, it is necessary to allocate at least 2 bits to the minimum temperature resolution. For example, when the temperature resolution of the device is 0.1 ° C, 0.02 per bit
The control accuracy of the FPN may be 0.025 ° C. at 5 ° C.

The block configurations shown in FIGS. 1 and 3 are block diagrams showing an example of the present invention, and it is of course possible to configure other circuits according to the gist of the present invention.

FIG. 5 is a block diagram showing the second embodiment according to the present invention. In the second embodiment, all the above-mentioned calculations are performed by the CPU 501, and the results are stored in the image memory 502 as needed. The program may be configured according to the flowchart of FIG. 503, 5
Reference numeral 04 is a ROM and a RAM for operating the CPU.

In the above description, as an estimation of the level of drift (xd), the following [Equation 3] is calculated from the displacement Δvi0, OB of each OB at the time of initialization and each OB displacement Δvi, OB in real time. By xd

[Equation 3] Are seeking. In addition to this estimation of xd, [Equation 4],
[Equation 5]

[Equation 4]

[Equation 5] Etc., there are several possibilities. [Formula 4] has a problem that miscalculation becomes large. [Formula 5] has a problem that calculation becomes complicated.

[0046]

As described above, according to the thermal infrared imaging device of the present invention, it is possible to correct the fixed pattern noise FPN generated due to the sensitivity variation of the thermal infrared detector, which was difficult to extend the prior art. Therefore, the temperature resolution can be further improved.

The temperature drift of each pixel includes not only the temperature drift of the device itself but also the temperature drift of the amplifier circuit. According to the present invention, this effect can be eliminated.

[Brief description of drawings]

FIG. 1 is a block diagram showing a first embodiment of a thermal infrared imaging device according to the present invention.

FIG. 2 is a diagram showing an operation of the thermal infrared imaging device according to the present invention.

FIG. 3 is a block configuration diagram further specifically showing the first embodiment of the thermal-type infrared imaging device according to the present invention shown in FIG.

FIG. 4 is a flowchart showing an embodiment of a thermal infrared imaging device according to the present invention.

FIG. 5 is a block configuration diagram showing a second embodiment of the thermal infrared imaging device according to the present invention.

FIG. 6 is a circuit configuration diagram showing a specific example of an infrared detector used in an embodiment of the thermal infrared imaging device according to the present invention.

FIG. 7 is a cross-sectional view of a conventional thermal infrared detection element.

FIG. 8 is a block diagram of a conventional infrared imaging device.

FIG. 9 is a diagram illustrating an operation of a conventional infrared imaging device.

[Explanation of symbols]

101 ... Infrared detector 102 ... Analog FPN correction circuit 103, 106, 107, 111 ... Subtractor 104 ... A / D converter 105 ... Digital FPN correction circuit 108 ... Scan converter 109 ... D / A converter 110 ... OB average value calculation circuit 112 ... OB displacement average value calculation circuit 113 ... Average value of OB drift coefficient 114 ... Divider 115 ... Multiplier 116 ... Drift coefficient memory 117 ... Drift correction circuit 118 ... Driving circuit 119 ... Peltier element 120 ... Peltier controller 301, 305 ... Adder 302, 306 ... Latch 303, 307 ... RAM 304, 308 ... Timing generator for OB part 309 ... Address generator 501 ... CPU 502 ... Image memory 503 ... ROM 504 ... RAM 601 ... Titanium bolometer 602 ... Pixel switch (N-type MOSFET) 603 ... Vertical AND 604 ... Transfer gate 605 ... Horizontal AND 606 ... Common source line 607 ... Vertical signal line 608 ... Horizontal signal line 609 ... Vertical shift register 610 ... Horizontal shift register 611 ... Output 612 ... Integrating circuit 613 ... Integrating transistor 614 ... Integral capacitance C1 615 ... Reset transistor 700 ... Semiconductor substrate 701 ... Scanning circuit 702 ... Silicon oxide film 703 ... Cavity 704, 705 ... Aluminum 706 ... Polysilicon 707 ... Titanium bolometer 708 ... Silicon oxide film 709 ... titanium nitride

─────────────────────────────────────────────────── ─── Continuation of front page (58) Fields surveyed (Int.Cl. ⁷ , DB name) G01J 1/02 G01J 1/42-1/44 G01J 5/02 G01J 5/12 G01J 5/20-5 / 26 G01J 5/48 G01V 9/04 H01L 27/14 H01L 37/00-37/02 H04N 5/30-5/335

Claims (2)

(57) [Claims] 1. A plurality of effective pixels having sensitivity to incident infrared rays and a plurality of reference pixels not having sensitivity to incident infrared rays, and the plurality of effective pixels and the plurality of effective pixels depending on a device temperature change. A temperature estimation device of a thermal infrared imaging device in which a signal level of a reference pixel changes, a means for obtaining a signal level variation among the plurality of reference pixels, and a signal level between the plurality of reference pixels at the time of initialization. Of the thermal infrared rays by dividing the variation of the signal level of the plurality of reference pixels, which is obtained by the means for obtaining the variation during normal operation, by the value stored in the memory. A temperature estimation device for a thermal infrared imaging device, which estimates the temperature of an imaging device. 2. A plurality of effective pixels having sensitivity to incident infrared rays and a plurality of reference pixels not having sensitivity to incident infrared rays, and the plurality of effective pixels and the plurality of effective pixels depending on a change in device temperature. In a drift estimation device for a thermal infrared imaging device in which the signal level of a reference pixel changes, a means for obtaining an average between the signal levels of the plurality of reference pixels, and a plurality of the average values obtained by the means for obtaining the average Of reference pixels and a means for obtaining the signal level displacement of the plurality of effective pixels and a memory for storing the same, and by changing the device temperature after FPN correction at the time of initial setting, the average value of each reference pixel and each effective pixel from A drift coefficient estimating device for a thermal infrared imaging device, characterized in that a displacement of a signal level is obtained , and the obtained value is stored in a memory as a drift coefficient of each pixel .