Image sensor circuit with signal processing function

Abstract

Unlike image sensors that perform processing within the pixel, this image sensor provides a mechanism that does not perform processing within the pixel, but instead integrates and holds signals outside the pixel within the sensor chip. [Solution] The image sensor circuit has three or more analog memories per pixel placed outside the pixels of the pixel array 11, sequentially and repeatedly accumulating the pixel signal in each, and reading out the result. Specifically, it comprises a pixel array 11 having L×M pixel groups arranged in a matrix, a memory array 15 having L×M×N memories with N (N≧3) analog memories placed outside the matrix to distribute the photocharge capacity of one pixel for each frame of the image sensor, an accumulator 13 that sequentially and repeatedly accumulates the photocharge capacity distributed to the N analog memories, and an output unit 17 that extracts the photocharge capacity accumulated from the memory array 15.

Description

[Technical Field]

[0001] This invention relates to an image sensor equipped with a signal processing function that converts electric field signals, such as electric field strength and phase, into modulated optical signals. [Background technology]

[0002] High-frequency electro-optic imaging involves converting an electric field signal into a modulated optical signal using electro-optic crystals, and then measuring this signal to obtain the electric field's intensity, phase, and distribution. Besides scanning individual measurement points, high-frequency electro-optic imaging can also be performed by simultaneously measuring multiple points using an image sensor. Simultaneous measurement of multiple points enables real-time imaging of the distribution. However, when measuring a large number of points simultaneously, the image sensor and signal detection / processing system are required to operate at extremely high speeds. This leads to factors such as larger imaging systems, increased power consumption, and higher costs. In reality, only a small portion of the vast amount of data is actually necessary, so by extracting only the necessary information at the initial stage whenever possible, the amount of data is reduced, resulting in smaller imaging systems, lower power consumption, and lower costs.

[0003] Conventional technology includes image sensors equipped with pixels that distribute the photocharge acquired by a photodiode within the pixel to different capacities (hereinafter referred to as lock-in pixels) (see Non-Patent Literature 1). Image sensors equipped with lock-in pixels can perform high-speed signal processing by manipulating and integrating the charge within the pixel. However, due to the limited area within each pixel and the small amount of charge that can be handled, it is unsuitable for handling weak changes and is not suitable for applications that require weak signal changes. Therefore, image sensors equipped with lock-in pixels are specialized for applications dealing with light that changes significantly, and Non-Patent Document 1 mainly focuses on distance measurement applications.

[0004] The present inventors have previously reported 10,000-pixel live electro-optic imaging (LEI) of electrical signals from high-frequency microwave circuits (see Non-Patent Literature 2). Non-Patent Literature 2 displays the near-field electrical field of a microwave circuit on a screen in real time at 30 frames per second. This report is based on photonics technology, applying the near-field electrical field to an electro-optic crystal plate and modulating the sensed light. The modulated frequency was down-converted using large-scale parallel processing photonic heterodyne, and the spatial pattern of the electro-optic modulation was detected by a high-speed image sensor. Furthermore, a digital signal processor (DSP) was used to extract the target frequency components and animate them on the screen. By observing fluctuations in the microwave electric field in real time, the operating state of the circuit can be intuitively analyzed and applied to diagnosis. The inventors further aimed to extend the LEI frequency from the microwave band to the millimeter-wave band, and applied modulated light generation technology to a camera to achieve real-time simultaneous multi-point measurement in the millimeter-wave band up to 100 GHz (see Non-Patent Literature 3).

[0005] In this type of high-frequency electro-optical imaging, the change in brightness is very small. Therefore, it is necessary to handle a large amount of charge and create conditions where the signal is larger than the noise caused by photon shot noise, etc. However, it is not practical to hold such a large amount of charge within a pixel, and reducing the area ratio of the photodiode that performs light detection within the pixel leads to a problem of decreased detection sensitivity.

[0006] Furthermore, image sensors capable of obtaining good images even in weak light environments are known (see Patent Document 1). This image sensor is a photon-counting type image sensor that counts the number of photons incident on a photodiode during the exposure period and outputs the photon count value as a signal value. In this photon-counting type image sensor, the counter overflows when it gets bright, so the counter is self-reset before it overflows. The timing of this self-reset can be either by resetting the entire memory array or by resetting the memory of individual pixels, but this self-reset concept has been applied to the memory of the photon-counting type image sensor. [Prior art documents] [Patent Documents]

[0007] [Patent Document 1] Japanese Patent Publication No. 2019-103121 [Non-patent literature]

[0008] [Non-Patent Document 1] Y. Shirakawa et. al, “An 8-Tap CMOS Lock-In Pixel Image Sensor for Short-Pulse Time-of-Flight Measurements”, Sensors, 20, 1040 (2020). [Non-Patent Document 2] K. Sasagawa et. Al., “Live electrooptic imaging system based on ultraparallel photonic heterodyne for microwave near-fields”, IEEE transactions on microwave theory and techniques, vol.55, issue 12, pp.2782-2791 (2007). [Non-Patent Document 3] K. Sasagawa et. Al., “Real-time digital signal processing for live electro-optic imaging”, Opt. Express, vol.17, no.18, pp.15641-15651 (2009). [Overview of the Initiative] [Problems that the invention aims to solve]

[0009] As mentioned above, in high-frequency electro-optic imaging, the amount of brightness change is very small, which presents challenges such as the need to handle a large number of charges and to make the signal larger than noise such as mechanical vibration noise and 1 / f noise. However, it is impractical to hold a large number of charges within a pixel, and it was also necessary to prevent a decrease in detection sensitivity without reducing the area ratio of the photodiode that performs light detection within the pixel.

[0010] In view of these circumstances, the present invention aims to provide an image sensor that differs from image sensors that process within pixels, by having a mechanism that does not perform processing within pixels but instead integrates and holds signals outside the pixels within the sensor chip. [Means for solving the problem]

[0011] To solve the above problems, the image sensor circuit of the present invention is characterized by having three or more analog memories per pixel outside the pixels of the image sensor, sequentially accumulating the pixel signal in each memory, and reading out the result. In the present invention, pixel signals are handled as voltages and accumulated in a capacitor outside the pixels of the image sensor, although within the same IC chip. Since the capacitor is arranged outside the pixels of the image sensor, a large capacitor that is difficult to arrange inside the pixels can be used. This makes it possible to handle more electric charge, and as a result, signals with weak changes can be captured. In addition, an integration circuit using an operational amplifier can be used to integrate pixel signals. This eliminates the need for a complex impurity implantation structure required for high-precision charge transfer within pixels, and can be implemented by a general standard CMOS process.

[0012] Specifically, the image sensor circuit of the present invention comprises: 1) an image sensor having an L×M (L≧2, M≧2) pixel group arranged in a matrix; 2) a memory array having L×M×N memories with N (N≧3) analog memories arranged outside the matrix for distributing the photocharge capacity of one pixel for each frame of the image sensor; 3) an integrating section that sequentially and repeatedly accumulates the photocharge capacity distributed to the N analog memories; and 4) an output section that extracts the accumulated photocharge capacity from the memory array. These are arranged on one chip to obtain an image sensor provided with a mechanism that integrates and holds pixel signals outside the pixels in the chip. The image sensor having an L×M pixel group is, for example, an image sensor with 100×100=10,000 pixels.

[0013] The image sensor is preferably a polarization image sensor. A polarization image sensor is used to observe objects that cannot be recognized by the human eye, such as electric field distribution, by capturing polarization (the vibration direction of light). The polarization image sensor is equipped with polarizers in multiple directions (for example, 0°, 45°, 90°, 135°), and can acquire polarization images in multiple directions in one shooting, thereby calculating the polarization direction and the degree of polarization.

[0014] For example, in the case of a polarization image sensor equipped with polarizers whose polarization directions of adjacent pixel pairs differ from each other by 90°, three or more analog memories per adjacent pixel pair can be arranged outside the pixels of the polarization image sensor, pixel signals can be sequentially and repeatedly accumulated in each analog memory, and the result can be read out.

[0015] An image sensor circuit using a polarization image sensor specifically comprises: 1) a polarization image sensor having L×M pixel groups arranged in a matrix; 2) a memory array having L×M×N / 2 memories, in which N (N≧3) analog memories for distributing the photocharge capacity of adjacent pixel pairs for each frame of the polarization image sensor are arranged outside the matrix; 3) an integrating section for sequentially and repeatedly accumulating the photocharge capacity distributed to the N analog memories; and 4) an output section for extracting the accumulated capacity from the memory array. Here, in an image sensor circuit using a polarization image sensor, for example, polarization pixels of 0° and 90° output complementary pixel signals, so integration can be performed with a phase shift of 180°, and the area of the memory array can be halved.

[0016] As will be described later, in the polarization image sensor, when there are N (N is an even number equal to or greater than 4) analog memories for storing photocharges of pixel pairs, and pixel signals of the pixel pairs have a phase difference of 90°, the pixel signals may be stored in the same analog memory.

[0017] The image sensor circuit of the present invention may further comprise an addition / subtraction section that performs addition and subtraction on pixel signals between analog memories corresponding to each pixel. Since the addition / subtraction section that performs addition and subtraction on pixel signals between analog memories corresponding to each pixel is provided separately from the integrating section, more advanced arithmetic processing can be performed within the chip.

[0018] In the image sensor circuit of the present invention, the integration unit may be composed of a switched-capacitor integration circuit. An integration circuit using an operational amplifier is used for integration, but this is configured as a switched-capacitor integration circuit. By using a switched-capacitor integration circuit and integrating and accumulating the difference between the input voltage and the reference voltage, the offset component can be reduced and the signal components can be integrated efficiently.

[0019] In the image sensor circuit of the present invention, it is preferable that the integration unit individually integrates the pixel signals of multiple frames to obtain two types of signals corresponding to the real and imaginary parts of the complex component, and performs some calculations of lock-in detection within the chip. Regarding the integration unit, the processing circuitry within the chip can be simplified by individually integrating the pixel signals of multiple frames.

[0020] A high-frequency electric field imaging device can be realized using the image sensor circuit of the present invention. This technology enables the detection of minute light changes at specific frequencies, which are required in applications such as high-frequency electric field imaging, to be performed faster and with a higher signal-to-noise ratio than simple image sensors. [Effects of the Invention]

[0021] According to the image sensor circuit of the present invention, by incorporating a mechanism that integrates and holds signals outside the pixel within the sensor chip without performing processing within the pixel, it is possible to handle more charges using a large capacitance that would be difficult to arrange within the pixel, and thus has the effect of being able to capture signals of weak changes. [Brief explanation of the drawing]

[0022] [Figure 1] Functional block diagram of the image sensor circuit [Figure 2] Schematic diagram of the image sensor circuit [Figure 3] Conceptual diagram of the measurement principle of electric field imaging. [Figure 4] Diagram illustrating electric field sensing [Figure 5]Schematic diagram of the LEI system [Figure 6] Diagram illustrating the calculation of lock-in detection. [Figure 7] Partial calculation flow diagram for lock-in detection [Figure 8] Diagram illustrating the integration process on the chip's memory (Example 1) [Figure 9] Diagram illustrating the integration process on the chip's memory (Example 2) [Figure 10] Functional block diagram of a conventional image sensor circuit [Figure 11] Conventional lock-in detection calculation process flow diagram [Modes for carrying out the invention]

[0023] Hereinafter, an example of an embodiment of the present invention will be described in detail with reference to the drawings. It should be noted that the scope of the present invention is not limited to the following embodiments or illustrated examples, and numerous modifications and variations are possible.

[0024] Figure 1 shows a functional block diagram of the image sensor of the present invention. As shown in Figure 1, the image sensor circuit 1 consists of a pixel array 11 which serves as the image sensor, an integration unit 13, a memory array 15, and a readout unit 17 which extracts signals from the memory array. Four analog memories are placed outside each pixel of the image sensor (pixel array 11), and the pixel signals are sequentially and repeatedly integrated into each of them, and the results are read out. Four arrows are drawn from the pixel array 11 to the integration unit 13, and from the integration unit 13 to the memory array 15. In reality, there are arrows corresponding to the number of pixels, and more specifically, arrows corresponding to the number of columns in the pixel array matrix.

[0025] Figure 2 shows a schematic diagram of the image sensor circuit. The image sensor circuit shown in Figure 2 is composed of a single chip. The pixel array 11 converts light incident on the pixels during the exposure period into electric charge using a photodiode and outputs a signal corresponding to the charge. The Y-scanner 12 selects multiple unit pixels in the pixel array 11 row by row using a switch and extracts the charge from the pixel array 11. The integration unit 13 integrates the extracted charge using an integration circuit with an operational amplifier and outputs it to a memory array 15 composed of analog memory. The charge held in each memory of the memory array 15 is controlled by the readout unit 17, which consists of an X-scanner 17a and a Y-scanner 17b, and is sequentially selected by the combination of the X-scanner 17a and the Y-scanner 17b.

[0026] Figure 3 shows a conceptual diagram of the measurement principle of electric field imaging. In electric field imaging, an electro-optic crystal 26 (ZnTe, LiNbO3, etc.) is used as a sensor to instantaneously measure the distribution of high-frequency electric fields, and the electric field distribution on the high-frequency circuit 22 is instantaneously visualized. An input signal 24 is applied to the high-frequency circuit 22, modulated laser light 20 is irradiated onto the electro-optic crystal 26, and the reflected laser light 21 is captured by an image sensor (not shown) to visualize the distribution of the nearby electric field. The electric field distribution image can obtain intensity (see Figure 3(1)) and phase (see Figure 3(2)).

[0027] Referring to Figure 4, the measurement principle of electric field sensing will be explained. In electric field sensing, the electric field signal is converted into an optical signal by the Pockels effect, which is then detected by an image sensor. An electro-optic crystal 26 is used as the electric field sensor. The electro-optic crystal 26 is used to measure the high-frequency near-field electric field of the device under test (DUT) on the high-frequency circuit 22. The measurement is performed by passing a laser beam back and forth through the electro-optic crystal 26 once, and detecting the change in polarization due to the change in birefringence caused by the application of an electric field to obtain electric field information. The birefringence characteristics are affected by the applied electric field, and the optical polarization of the laser beam 21 passing through the electro-optic crystal 26 is modulated by the electric field 27 and converted into intensity modulation. The output signal is detected by a photodiode. In the LEI system, all sample points are measured in parallel by optical image sensors, significantly reducing image acquisition time and enabling real-time image acquisition at video rates.

[0028] Figure 5 shows a schematic diagram of the LEI system. The image sensor 11 has 10,000 pixels and a frame rate of 20 kHz. The laser light source 31 uses a 780 nm single-mode laser diode, which is highly sensitive to the Si-based image sensor 11. The reference signal from the image sensor 11 is supplied to the signal generator. Since the frame rate of the image sensor 11 is much lower than the RF frequency of the electric field, the RF signal is down-converted by the optical heterodyne method. The continuous wave laser light emitted from the laser light source 31 has a frequency f LO The light is modulated by a Mach-Zehnder type optical modulator 32 and output from probe 33. The polarization of the light is determined by the measured electric field frequency f RF The signal is then modulated again within the electro-optic crystal 26. The typical power of the modulated light is approximately 4 mW. The polarization modulation is converted to intensity modulation by the polarization beam splitter 37. The waveplate is adjusted to obtain the best sensitivity. The output light contains an intermediate frequency component f that reflects the intensity and phase of the electric field signal. IF =|f RF -f LO | is included. Here, f LO is, f IF The frequency is selected to be lower than the Nyquist frequency of the image sensor 11. In the LEI system, f IF The frequency is set to 5kHz, and the intermediate frequency is one-quarter of the frame rate of the image sensor 11. Here, the reference signal is generated by the image sensor circuit 1 and input to the signal generator 41. The signals detected by each pixel of the image sensor 11 are calculated by the integration unit 13 on the same chip and stored in the memory array 15. The data is read out from the memory array 15 by the readout unit 17 and transferred from the image sensor circuit 1 to the PC (personal computer) 19, and the electric field image is displayed on the PC 19. Furthermore, the electric field 27 generated by the object under test (DUT) 25, which is supplied with a signal from the signal generator 42, modulates the optical polarization of the laser light that has passed through the electro-optic crystal 26, and the output signal is detected by the photodiode of the pixel in the image sensor 11. In Figure 5, reference numerals 34 and 40 indicate lenses, reference numerals 35, 36, 38, and 39 indicate waveplates, and reference numeral 43 indicates a mirror.

[0029] Figure 6 shows an explanatory diagram of the lock-in detection calculation. A key advantage of the LEI system is high-speed image acquisition and real-time imaging. To achieve this, it's necessary to simultaneously acquire electric field signals from each pixel of the image sensor. The data is first processed by a DSP and filtered to reduce data size. Next, images of the electric field amplitude and phase are extracted, sensitivity and phase artifacts are corrected on a PC, and the image is displayed.

[0030] As mentioned above, the inventors have demonstrated high-speed imaging of electric fields up to the millimeter wave band by combining a weakly polarized image sensor and an electro-optic crystal. However, since the frequency of the observed electric field is high relative to the image sensor's frame rate, it is down-converted to an intermediate frequency by optical heterodyne detection. For example, the frame rate is set to 360 FPS and the intermediate frequency to 90 Hz. However, this frequency band is susceptible to mechanical noise and other influences, making it necessary to improve the intermediate frequency for higher sensitivity. This requires increasing the image sensor's frame rate, but as a result, the load on data transfer from the chip to the PC and signal processing, including lock-in detection, becomes a problem. To solve this problem, a circuit was designed that processes part of the lock-in detection within the image sensor chip, thereby reducing the load on subsequent processing. In other words, the data processing performed by the DSP on the signal output from the image sensor 11 (the calculation logic for lock-in detection shown in Figure 6) was simplified by changing the calculation order in the integration unit 13.

[0031] FIG. 7 is a processing flow of performing a part of lock-in detection calculation in a chip of an image sensor circuit. First, the intermediate frequency is set to 1 / 4 of the frame rate through optical heterodyne detection (step S01). Four frames (i1, i2, i3, i4) are regarded as one set (step S02), and the four-frame signals (photoelectric charges) are individually integrated inside the chip (step S03). Specifically, the calculation shown in step S04 is performed, data is transmitted from the inside of the chip to an external PC (step S05), and intensity and phase are calculated by the PC (step S06). Details will be described below.

[0032] FIG. 10 is a functional block diagram of a conventional image sensor circuit, and FIG. 11 shows a calculation processing flow of conventional lock-in detection. In the conventional image sensor circuit, one memory is allocated per pixel, and processing of the integrating unit 103 is data processing performed by a DSP, so the circuit is complicated. As shown in the conventional lock-in detection calculation processing flow of FIG. 11, the intermediate frequency is set to 1 / 4 of the frame rate, four frames (i1, i2, i3, i4) are taken as one set, and a DSP outside the chip uses the following equation 1 to obtain v cos , v sin ​.

[0033] [Math.]]

[0034] By means of this addition and subtraction, the same result is obtained as obtaining a rectangular wave signal with a phase difference of π / 2 and multiplying it by each signal, which respectively correspond to the real part and the imaginary part of the complex component. Then, the DSP uses the following equation 2 to integrate the two types of signals to obtain V cos , V sin ​.

[0035] [Math.]]

[0036] In equation 2 above, n is the period number and N is the number of integrated frames. This calculation functions as a low-pass filter, so the output was filtered to retain only the DC component. Using a PC or similar device, the intensity V was determined from this result. cos 2 +V sin 2 and phase arctan(V sin / V cos By calculating ), we were able to obtain the electric field strength and phase image (intermediate frequency component) using only simple addition and subtraction. In this calculation, further improving the frame rate requires high-speed performance from the AD converter, while the slow transfer time to the PC becomes a problem. The inventors noticed that the result does not change even if the above calculation order is reversed, and reduced the amount of data to be transferred by integrating i1 to i4 first within the chip, thereby reducing the performance requirements for the subsequent AD converter. Specifically, as shown in Equation 3 below, (Σi1,··,Σi4) was obtained by individually integrating the signals of the four frames within the chip. Then, the intensity and phase were determined by transferring this data to a PC and calculating using Equation 4 below.

[0037]

number

[0038] JPEG2026148362000005.jpg16148

[0039] In fact, to perform the calculation in equation 3 above, we fabricated an image sensor circuit that can separate the frame into four parts and integrate the signal from each pixel. Conventional field imaging methods involve irradiating electro-optic crystals with high light intensity to read out minute birefringence changes, and acquiring signals under conditions where photon shot noise becomes dominant. This requires extremely large capacity to store multiple frame signals. However, it is difficult to provide multiple such large pixel capacities within a single pixel. Therefore, a capacitor for integrating and storing the signal voltage is placed outside the pixel, allowing for a sufficiently large capacity. In other words, since there is not much space inside the pixel, the capacitor capacity is secured outside the pixel to handle the charge sufficiently. By placing the memory capacity outside the pixel, it becomes possible to increase the capacity value by more than 10 times, for example. Since the upper limit of the signal-to-noise ratio caused by shot noise is determined by the square root of the number of charges being handled, it is estimated that the signal-to-noise ratio improves by approximately 3.3 times when handling 10 times the number of charges. In particular, in high-frequency electric field imaging, the rate of change of the signal is low, so this improvement in the signal-to-noise ratio also improves the signal-to-noise ratio of the acquired electric field signal. Furthermore, a switched-capacitor type integrating circuit is used to integrate and store the difference between the input voltage and the reference voltage. This reduces the offset component and allows for efficient integration of signal components. As a result, it becomes possible to integrate a large number of frames and measure weak signals. [Examples]

[0040] The integration process on the in-chip memory of the image sensor circuit in Example 1 will be explained with reference to Figure 8. As shown in Figure 8, there are four analog memories (hereinafter simply referred to as "memory") per pixel, and the pixel signal is sequentially and repeatedly accumulated in each memory. This reduces the number of data transfers and enables an improvement in the frame rate. The accumulated results can be read out sequentially at any desired timing. This type of image sensor circuit enables lock-in detection preprocessing within the pixel, allowing for high-speed modulation and reducing low-frequency noise. [Examples]

[0041] The integration process on the in-chip memory of the image sensor circuit in Example 2 will be explained with reference to Figure 9. As shown in Figure 9, the two types of polarized pixels, 0° and 90°, output complementary pixel signals, allowing them to be integrated with a 180° phase shift. Because addition is performed and two types of polarized pixels are used, if the timing is right, calculations can be performed with just four memory units for a pair of pixels. Since the waveforms of two pixel pairs are the same except for a 180° phase difference, the timing can be changed to duplicate the use of four memory units per pixel. [Industrial applicability]

[0042] This invention is useful for high-frequency electric field imaging, where the detection of minute changes in light is required. [Explanation of Symbols]

[0043] 1. Image sensor circuit 11-pixel array (image sensor) 12,17b Y-scanner 13,103 Estimation Department 15,105 memory arrays 17 Readout section 17a X-scanner 19 PC 20 Modulated laser light 21 Reflected laser light 22 High-Frequency Circuits 24 Input Signals 25 Object to be measured 26 Electro-optic crystals 27. Electric field 31 Laser light source 32. Mach-Zehnder type optical modulator 33 probes 34,40 lenses 35,36,38,39 Wave plate 37 Polarizing Beam Splitter 41 Signal Generator 42 Signal generator 43 Miller

Claims

1. An image sensor circuit that has three or more analog memories placed outside each pixel of the image sensor, sequentially and repeatedly accumulating the pixel signal in each memory, and reading out the result.

2. An image sensor having an L×M pixel group arranged in a matrix, A memory array having L × M × N memories, wherein N (N ≥ 3) analog memories for distributing the photocharge capacitance of one pixel for each frame of the image sensor are arranged outside the matrix, An integration unit that sequentially and repeatedly accumulates the photoelectric charge capacities distributed to the N analog memories, An output unit that extracts the photoelectric charge capacity stored from the memory array, The image sensor circuit according to claim 1, comprising:

3. The image sensor circuit according to claim 1, wherein the image sensor is a polarization image sensor.

4. A polarization image sensor equipped with polarizers whose polarization directions differ by 90° from each other, wherein three or more analog memories are placed outside the pixels of the polarization image sensor for each adjacent pixel pair, and the image sensor circuit is capable of sequentially and repeatedly integrating the pixel signals into each and reading out the results.

5. A polarization image sensor having an L×M pixel group arranged in a matrix, A memory array having L × M × N / 2 memories, in which N (N ≥ 3) analog memories are arranged outside the matrix for distributing the photocharge capacitance of adjacent pixel pairs for each frame of the polarization image sensor, An integration unit that sequentially and repeatedly accumulates the photoelectric charge capacities distributed to the N analog memories, An output unit that extracts the capacity stored from the memory array, The image sensor circuit according to claim 4, comprising:

6. The polarization image sensor is The analog memory for accumulating the photocharge of the aforementioned pixel pair is N (an even number such that N ≥ 4), The image sensor circuit according to claim 5, wherein the pixel signals of the pixel pair are stored in the same analog memory when they have a phase difference of 90°.

7. The image sensor circuit according to any one of claims 1, 2, 4 to 6, further comprising an addition / subtraction unit for adding and subtracting pixel signals between the analog memory corresponding to each pixel.

8. The image sensor circuit according to any one of claims 2, 5, or 6, wherein the integration unit is composed of a switched-capacitor integration circuit.

9. The image sensor circuit according to any one of claims 2, 5, or 6, wherein the integration unit individually integrates the pixel signals of multiple frames to obtain two types of signals corresponding to the real and imaginary parts of the complex component, and performs some calculations of lock-in detection within the chip.

10. A high-frequency electric field imaging apparatus using an image sensor circuit according to any one of claims 1, 2, 4 to 6.

Images