JPH0638110A - Contour detection system - Google Patents

Abstract

PURPOSE:To provide the contour detection system realizing the contour detection in the picture processing technology in real time without scanning of a bias voltage. CONSTITUTION:A part surrounded by broken lines is one unit of circuit and the system is provided with two semiconductor light receiving elements 1,2 whose electrode structure is symmetrical to the left and right, a terminal 3 outputting an electric signal, and two wires 4, 5 receiving a bias voltage. Plural circuits C11-Cnn are arranged in two-dimension to form one contour detection system. Bias voltages +Vb, -Vb opposite in the polarity and equal in the amplitude are applied in common in advance to each longitudinal column so that unit circuits adjacent to each other have a common bias line. When no optical input is given or an optical input is given to both the light receiving elements 1, 2, no signal appears at the output terminal 3. When an input picture has a contour just on the unit circuit, a difference of the light intensity appears at the output terminal 3.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a contour detection system in image processing technology.

[0002]

2. Description of the Related Art A semiconductor light receiving element having a bilaterally symmetric electrode structure is used and two-dimensionally arranged on a semiconductor substrate so that the contour of an input image can be detected. In the prior art, as a semiconductor light receiving element, it is generally an MS.
It is realized by using an element having a comb-shaped electrode structure in which Schottky junctions are connected in opposite directions in a light receiving portion, which is known as an M photodetector.

In this configuration, each light receiving element functions as an independent diode, and these are connected in a two-dimensional matrix. These light receiving elements are arranged in a grid pattern as shown in FIG. Then, the bias voltage is applied in the vertical row, and the output current is read in the horizontal row. That is, + 2V is applied to the symmetrical contour detection sequence,
Apply a bias voltage of -1V to the columns on both sides of this,
An electric signal is input only when there is a difference in brightness between input images on this row. By scanning this bias voltage pattern over all the electrode columns, vertical contour detection can be performed.

This technique is described in "1992 Spring Applied Physics Association Joint Lecture Proceedings 29a-B-5".

[0005]

DISCLOSURE OF THE INVENTION In the prior art,
Bias voltage must be scanned to detect contours, and to detect contours in both vertical and horizontal directions,
It was necessary to change the direction and scan again. In addition, there is a drawback that the waveform of the bias voltage to be scanned is complicated. In particular, in this case, since there is no function of accumulating incident light on the circuit, the sensitivity cannot be increased by this scanning, and there is no particular advantage due to scanning.

Further, the MSM photodetector is originally
Although it is a light-receiving element characterized by having a picosecond-order ultra-high-speed response characteristic, in the conventional technology, the bias voltage pattern must be scanned over all electrode rows,
There has been a problem that the processing time per screen becomes very long. Especially, when the number of pixels is large, this is a big drawback, and the characteristics of the light receiving element have not been utilized. On the other hand, there has been a demand for a device capable of statically detecting the contour without scanning the bias voltage.

In view of the above problems, the present invention provides a method for realizing a contour detecting operation performed by a light receiving element arranged two-dimensionally on a semiconductor substrate in real time without scanning a bias voltage. It is a thing.

[0008]

In order to solve the above problems, a photodetector having an electrode structure in which a pair of electrodes are opposed to each other on the left and right to keep a dark current small in a substantially symmetrical direction is provided on a semiconductor substrate. A point provided and electrically connected to the two light receiving elements is used as an electric signal output terminal, and bias voltages having opposite polarities and equal in magnitude to each other when viewed from the electric signal output terminal are applied to the two light receiving elements. One unit circuit is a circuit that outputs an electric signal from the electric signal output terminal only when light is input to either one of the two light receiving elements. A plurality of unit circuits are arranged on the semiconductor substrate. It is characterized in that it integrates and detects the contour of the input image pattern.

Further, in this adjacent unit circuit, each light receiving element may be arranged so that the light receiving regions thereof overlap with each other, so that the contour non-detection portion does not occur. Further, a plurality of unit circuits in which light receiving elements are arranged and connected in the vertical, horizontal, right diagonal, and left diagonal directions are arranged adjacent to each other to form a unit group, and a plurality of unit groups are arranged. Thus, the contour of the input image pattern in four directions can be detected.

Further, it may be characterized in that the contour of the color or wavelength of the inputted image pattern can be detected by forming the set of light receiving elements in the unit circuit so as to have different response wavelength ranges. The output terminal may be complementarily connected to a circuit having a low-pass filter characteristic in a time series or space series, or an input terminal of a set of semiconductor light emitting elements. The light-receiving element may have a structure in which rectifying junctions are connected in opposite directions, and the rectifying junction may be a Schottky junction.

[0011]

According to the present invention, when there is no light input, no signal appears at the output terminal, and even when two light receiving elements in the same unit circuit have the same light input, both signals are received. Since bias voltages having opposite characteristics to each other when viewed from the output terminal and having the same magnitude are applied, the output becomes zero because the photocurrents from both sides cancel each other out, and no signal appears at the output terminal. On the other hand, when the input image has a contour just on this unit circuit, that is, when the light intensities incident on both of the two light receiving elements are different, the difference between the outputs of the two light receiving elements is the output terminal. Appear in. By arranging a plurality of unit circuits two-dimensionally, it is possible to detect the contour of the input image.

With the system according to the invention, it is possible to perform image processing at a very high speed, since it is not necessary to scan the bias voltage across the array of transfer electrodes. Specifically, detection can be performed at a speed equal to the response speed of the light receiving element.

[0013]

EXAMPLES As the semiconductor element used in the present invention, a planar type element having a substantially symmetrical electrode structure is used. A typical example thereof is a photoconductor formed by facing ohmic contact electrodes on a semiconductor substrate. This has a problem that if the inter-electrode resistance is increased in order to suppress the dark current to be low, the operation speed becomes slow, but since the current output (gain) at the time of light input is large, the LE is connected to the output terminal.
When connecting a semiconductor light emitting device such as a D or a semiconductor laser (LD), there is an advantage that they can be directly driven without an amplifier circuit.

Further, for ultra-high speed operation, an MSM made by facing Schottky junction electrodes on a semiconductor substrate.
A photo detector is suitable. This is effective for an extremely dark input image because the dark current is small, although the LED or LD cannot be directly driven.

A semiconductor substrate in which two light receiving elements are set as one set, a point where these light receiving elements are connected to each other is used as an electric output terminal, and a circuit to which bias voltages having opposite polarities are applied to both sides is defined as one unit It can be realized by arranging a plurality of them on top and integrating them.

Next, a detailed description will be given while showing a concrete structure. FIG. 1 shows a basic circuit configuration of a contour detection system according to the present invention. The part surrounded by the broken line is 1
It is a unit circuit and is composed of two semiconductor light receiving elements 1 and 2 having a symmetrical electrode structure, a terminal 3 for outputting an electric signal, and two wirings 4 and 5 to which a bias voltage is applied. A plurality of these (C _{11 to} C _nn ) are arranged two-dimensionally to form one contour detection system. The columns in each longitudinal, with opposite polarities to each advance one row equals the bias voltage + V _b of magnitude, -V _b are commonly applied, the unit circuit of next to each other to share a common bias line It is like this.

The operation of one unit circuit is as follows.
First, when there is no light input, no signal appears at the output terminal 3. Even when light is input to both of the two light receiving elements 1 and 2, since bias voltages having opposite characteristics when viewed from the output terminals and having the same magnitude are applied to both, the photocurrents from both are canceled. The output is zero to fit.

On the other hand, when the input image has a contour just on this unit circuit, that is, when the light is incident on only one of the light receiving elements or the light intensity incident on both is different, The difference appears at the output terminal 3. In this case, when the light intensity input to the light receiving element 1 to which the positive bias voltage is applied is higher, the electrode 3
A positive electric signal appears at. On the contrary, when the light intensity input to the light receiving element 2 to which the negative bias voltage is applied is higher, the negative electric signal appears on the electrode 3.

By arranging a plurality of unit circuits two-dimensionally, it is possible to detect the contour of the input image. The detection result is output as an electric signal from the output terminal 3 for each unit circuit (pixel). , Output independently.

In this case, unlike the prior art, it is not necessary to scan the bias voltage over the entire electrode array, so that image processing can be performed at a very high speed. In particular,
Since detection can be performed at a speed equal to the response speed of the light receiving element, when the MSM photodetector is used,
At the output terminal 3, full output can be obtained in an extremely short time on the order of picoseconds. These output signals are again displayed as an image on the display through the signal processing processor connected to the subsequent stage, or further processed. Alternatively, as described later, it is possible to connect and integrate semiconductor light emitting elements such as LEDs and LDs so that the contour is immediately output as an image pattern from the same surface of the semiconductor substrate or the opposite surface. .

The light receiving element used here is required to have a small dark current in both positive and negative polarities in order to improve the S / N ratio. Those satisfying this condition are preferably a photoconductive element such as a CdS cell, or an element such as an MSM photodetector in which rectifying junctions are connected in opposite directions.

As the semiconductor substrate used here, Si or GaAs or CdS is used when the input image is visible light, and Ge or InG is used when the input image is near infrared light.
In the case where aAs is far infrared light, InSb, Hg
CdTe or the like is used.

The contour detection system described above can detect contours only in columns in the vertical direction, and has no detection capability in other directions. Therefore, FIG. 2 shows an embodiment in which contour detection can be performed simultaneously in the vertical and horizontal directions, and this will be described.

The unit circuits surrounded by broken lines are arranged two-dimensionally in both the vertical and horizontal directions as shown in the figure, and the bias lines are obliquely provided in order to alternately apply voltages of opposite polarities to the respective bias points. Wire. As a result, contour detection can be performed simultaneously in the vertical and horizontal directions. In this case, there are various methods of arranging the light receiving elements and wirings thereof, and the arrangement is not limited to that shown in FIG. For example, as shown in FIG. 3, the wiring of the bias line is somewhat complicated, but the light-receiving elements can be arranged at finer intervals, and the resolution of image processing can be improved.

Further, in FIG. 4, the light receiving elements can be arranged more precisely than in FIGS. 2 and 3, and contours in both vertical and horizontal directions are detected. In this configuration, the vertical light receiving elements and the horizontal light receiving elements are arranged alternately, and the output side is provided like crossing.

Since the contour detecting circuit shown above is actually constructed by integrating and arranging a planar type light receiving element having a simple left-right symmetric structure on a semiconductor substrate, the original resolution is very high. However, for a finer input screen, there is a possibility that a defect occurs because the contour detection function is not provided at the boundary line where the unit circuits are adjacent to each other.

In order to compensate for this, FIG. 5 shows an arrangement in which adjacent light receiving elements are arranged so that their light receiving regions overlap each other. That is, the paired light receiving elements are arranged so as to be sandwiched between the respective one light receiving elements of the adjacent left and right unit circuits. As a result, when the contour portion of the input image is specifically located at the portion A in the figure, the contour is detected over the two unit circuits, and the contour is B
When moving to the part (1), no signal is output from the left unit circuit, so that the contour is detected from only one unit circuit. Further, when the contour moves to the portion C, this time, a signal can be obtained from the unit circuit on the right side as well. In this way, it is possible to detect the outline of the input image between any of the light receiving elements.

FIG. 6 also has the same function for the vertical contour, and is obtained by superimposing the one obtained by rotating the entire body by 90 degrees in FIG. As a result, it is possible to detect contours in both vertical and horizontal directions without insensitive parts.

FIG. 7 corresponds to another embodiment of the example of FIG. 6, in which the light receiving elements are more precisely arranged while the contours in both the vertical and horizontal directions are prevented from generating non-detection lines. Here, as in FIG. 4, the vertical light receiving elements and the horizontal light receiving elements are arranged alternately. The vertical and horizontal arrangements that were straight lines in FIG. 6 are zigzag here, and the light receiving elements are wired so that the light receiving regions of the adjacent unit circuits overlap each other as shown in FIGS. 5 and 6. By doing so, the light receiving elements can be arranged more precisely.

FIGS. 13 and 14 are conceptually easy-to-understand diagrams showing the connection of the light-receiving elements in the systems of FIGS. 6 and 7. In these figures, white circles indicate light receiving elements.

In addition to these, FIG. 8 shows an embodiment in which contour detection can be simultaneously performed in the oblique direction. C _ij -1~C _ij -4 represents the unit circuit similar to that described above, they are respectively 4
Those arranged so as to be adjacent to each other in the direction are set as one group unit and are shown by a chain line. By arranging these unit groups two-dimensionally, it is possible to simultaneously detect contours in four directions of vertical / horizontal / diagonal right / diagonal left. Therefore, pattern detection can be performed for virtually any input image. It will be possible.

Further, the wavelength of the input image, that is, the contour of the color can be detected by providing a difference in the response wavelength range between the two light receiving elements in the unit circuit. For this, there are a method of providing different color filters on the two light receiving elements and a method of forming the two light receiving elements using different semiconductor materials. The latter is realized by forming light absorption layers made of different materials corresponding to respective light receiving elements on the same semiconductor substrate. For this, a method of forming a different semiconductor growth layer such as (In) GaAs or Ge on a Si substrate, or a 3- or 4-element III-
There is a method of adjusting the mixing ratio of elements using a group V compound semiconductor, for example, a method of forming Ga _1-x Al _x As so that the x values thereof are different from each other.

Further, a group in which the combination of the response wavelength ranges is changed between adjacent unit circuits is made into one group unit, and this is arranged two-dimensionally so that a plurality of wavelengths or colors can be displayed on the input screen. Contour detection is also possible. Specifically, it can be realized by providing a plurality of different color filters in a mosaic pattern on the light receiving element like a color CCD. According to this, since the contour for each wavelength or color is obtained, the color separation of the input image is performed. For example, by displaying this result on a monitor and giving an instruction to fill the inside of the contour with the color from an external circuit, a kind of color imaging becomes possible. Further, it is also possible to output the obtained contour pattern signal to an external circuit and perform various processes for each wavelength.

Next, regarding a method for connecting a circuit composed of a semiconductor light emitting element such as an LED or an LD to the output terminal in order to output the contour detection result as an image pattern and integrating the circuit, for one unit circuit A description will be given while showing a specific example.

In FIGS. 9A and 9B, the output electric signal is amplified by connecting the gate of the field effect transistor (FET) to the output terminal and passing the gate through the gate. In the circuit configuration of the present invention, since an electric signal of both positive and negative polarities is output from the output terminal, as shown in FIG.
It is necessary to connect both the channel and p-channel FETs 7, 8. Further, since the semiconductor light receiving element generally has a high impedance and the output is taken out as a current signal, it is desirable for high speed operation that the output terminal is installed via a low resistance 6 such as 50 ohms.

As shown in FIG. 9A, the drain of the FET =
By connecting the semiconductor light emitting elements 9 and 10 in series to the source circuit, the detected contour can be output as an image pattern. Further, as shown in FIG. 9B, if the drain = source circuit of the FET is connected in parallel, an inverted image pattern in which the detected contour is shown as “dark” can be obtained. Here, 11 and 12 are load resistors on the output side of the FET.

As the amplification element connected to the output terminal 3, a bipolar transistor may be used instead of the FET. In this case, since the input of the transistor is generally a current signal, it is not necessary to install the output terminal 3 via a resistor.

FIG. 10 shows a method for directly connecting the semiconductor light emitting element to the output terminal. When a photoconductive element composed of an ohmic contact such as a CdS cell is used as the semiconductor light receiving element of the light input section, a large current output can be obtained, so that the light emitting element can be directly driven. In this case as well, it is necessary to complementarily connect the two light emitting elements 9 and 10 so as to correspond to both the positive and negative directions. It should be noted that the circuit connected to the output terminal shown here is only an example thereof, and if it is modified to a complementary circuit configuration, it is generally a semiconductor such as an impedance conversion circuit. All circuits connected to the subsequent stage of the light receiving element can be connected here.

Here, the semiconductor substrate used is
Si for integration including light emitting devices and transistors
Or GaAs or InP is used. Of these, S
Although i is suitable for a transistor, it cannot be used to directly form a light emitting device. However, at present, it is possible to form a direct transition type semiconductor crystal such as GaAs on a Si substrate by an epitaxial growth method, which makes it possible to integrate these semiconductor elements together on the same Si substrate. Is.

As described above, as shown in FIG. 9 and FIG.
It is necessary to connect the circuit of the next stage to the output terminal of this circuit so as to correspond to the electric output signals of both positive and negative polarities, which causes a problem that the circuit becomes somewhat complicated. However, as an advantage in this case, by checking the polarity of the electric signal to be output or, when the light emitting elements are integrated, by checking which light emitting element has emitted light, It is possible to detect the direction of the change in light and darkness at, that is, the polarity of the gradient of the light input intensity. Further, since it is equivalent to being able to extract the differential signal of the contour signal in that row, it is possible to detect the contour with substantially double the resolution.

Further, as a matter of course, the light receiving element and the transistor or the light emitting element used here all have an output that linearly changes with respect to an input signal, so that a threshold value is provided to digitally In addition to detecting only the presence or absence of contours, the degree of lightness and darkness may be detected in an analog manner. In this case, it is possible to add an ambiguous characteristic that is closer to the human recognition operation, and this may be applied to a neurodevice or the like.

A feature of the present invention is that the contour output can be two-dimensionally output even under a static bias voltage, so that scanning is not required. However, as shown in FIG. Can also be scanned. That is, by separately providing the shift registers 13 and 14 for scanning the bias voltages of both positive and negative polarities and shifting both of them with good timing so that the bias voltages of both polarities are always applied to the unit circuit, Done.
The output signals are output circuits 15-1 to 15-1 provided for each column.
15-n, and further sequentially output to the outside in time series over all columns. This is effective when the contour detection result is input to an external arithmetic processing device such as a computer.

Further, in this case, since the output contour pattern is extracted as a time-series signal, it is possible to perform some processing by inputting this output signal to the frequency filter circuit. Generally, in an operation such as contour extraction in which a difference between pixels is obtained, the operation is vulnerable to spatial noise and has a defect that a flaw or the like of an input image is further emphasized and output. Therefore, by performing a process of inputting a time-series output signal to a filter circuit (for example, an integrating circuit) that cuts off high frequencies, it is possible to remove extremely short line segments such as scratches on an image. Enables efficient criticality detection.

FIG. 11 shows an example in which the embodiment shown in FIG. 1 is improved. However, an element capable of detecting a contour in a plurality of directions as shown in FIGS. The same can be applied to a device in which a contour non-detection portion is removed by arranging light receiving elements between circuits so that regions overlap each other, and further to an element for detecting a contour of a wavelength described above.

In the case of an element having an FET connected to the output terminal, the bias voltage (drain voltage) of the FET can be scanned by statically applying the bias voltage of the light receiving element. In this case, by increasing the value of the load resistance of the output terminal 3 shown in FIG. 9, the output terminal of the light receiving element, that is, the input terminal of the FET has a high impedance, so that an accumulation action of the output signal occurs and the high sensitivity Can be detected.

According to the invention described above, a semiconductor light receiving element having a substantially symmetrical electrode structure is used, and a plurality of the semiconductor light receiving elements are two-dimensionally arranged on a semiconductor substrate to be integrated. The contour of the input image can be detected at an extremely high speed of the order. Further, since the semiconductor device is manufactured by a normal semiconductor element manufacturing technique, the advantages of the normal semiconductor integrated circuit such as small size, good resolution and low cost can be directly applied to the present invention.

[0047]

As described above, by using the contour detection system and the element according to the present invention, it becomes possible to realize an ultra-high-speed image recognition function, which is impossible in the present computer, and to realize a real-time operation for a complicated input screen. Great progress can be made towards the probability of image processing technology. In particular, the present invention is very effective as a preprocessing processor at the image input stage, that is, as a biological retina function, in order to use a computer to perform the same function as a human image processing function.

[Brief description of drawings]

FIG. 1 is a schematic diagram of a basic contour detection system in an embodiment of the present invention.

FIG. 2 is a schematic diagram of a contour detection system that is capable of simultaneously detecting contours in both vertical and horizontal directions.

FIG. 3 is a schematic diagram of a contour detection system that is capable of simultaneously detecting contours in both vertical and horizontal directions.

FIG. 4 is a modification of FIGS. 2 and 3, and is a schematic diagram of a contour detection system capable of detecting contours in both vertical and horizontal directions.

FIG. 5 is a schematic diagram of a contour detection system in which a portion where contour detection cannot be performed is removed by arranging light receiving elements between adjacent unit circuits so that regions overlap each other.

FIG. 6 is a schematic diagram of a contour detection system configured to detect contours in both vertical and horizontal directions simultaneously from FIG. 5;

FIG. 7 is a schematic view of the contour detection system in which the light receiving elements are arranged more precisely from FIG. 6;

FIG. 8 is a schematic diagram of a unit group of a contour detection system that is capable of simultaneously detecting contours in four directions of vertical / horizontal / right diagonal / left diagonal.

FIG. 9 is a diagram of a unit circuit in which a semiconductor light emitting element is connected to an output terminal via a transistor.

FIG. 10 is a diagram of a unit circuit in which a semiconductor light emitting element is directly connected to an output terminal.

FIG. 11 is a schematic diagram of a contour detection system configured to scan a bias voltage of a light receiving element and obtain an output signal in time series.

FIG. 12 is a diagram showing a circuit of a contour detection system in the related art.

13 is a diagram conceptually showing the connection between the light receiving elements of the system in FIG.

14 is a view conceptually showing the connection between the light receiving elements of the system in FIG.

[Explanation of symbols]

1, 2 ... A semiconductor light receiving element having a substantially symmetrical electrode structure in which a pair of electrodes are provided on a semiconductor substrate so as to face each other, 3 ...
Electrical signal output terminal, 4 ... Wiring to which a positive bias voltage is applied, 5 ... Wiring to which a negative bias voltage is applied, 6 ... Load resistance of the output terminal 3, 7 ... Field effect transistor (n channel) , 8 ... Field effect transistor (p channel), 9, 10 ... Semiconductor light emitting element, 11, 1
2 ... Load resistance on transistor output side, 13 ... Shift register for scanning positive bias voltage, 14 ... Shift register for scanning negative bias voltage, 15-1 to 15-n ... Output circuit of element for scanning bias voltage, C ₁₁ ~ C _nn , C _ij -1 ~
C _ij -4 ... 1 unit of contour detection circuit, + V _b , −V _b ... Bias voltage, 101 ... MSM photodetector two-dimensional array, 102 ... Bias voltage shift circuit, 103 ... Current output terminal array.

Front page continued (72) Inventor Takashi Iida 1 1126, Nomachi, Hamamatsu, Shizuoka Prefecture 1126 Hamamatsu Photonics Co., Ltd. (72) Inventor Sadahisa Warashi 1 1126, Nonomachi, Hamamatsu, Shizuoka Prefecture 1 Hamamatsu Photonics Co., Ltd. (72) Inventor Kenichi Sugimoto, 1 Hamamatsu Photonics Co., Ltd., 1126, Nomachi, Hamamatsu, Shizuoka Prefecture (72) Inventor Tomoko Suzuki 1 1126, Ichinocho, Hamamatsu, Shizuoka (1) In Hamamatsu Photonics Co., Ltd. (72) Invention Hirofumi Suga 1 Hamamatsu Photonics Co., Ltd. 1 No. 1126 Nomachi, Hamamatsu City, Shizuoka Prefecture

Claims (8)

[Claims] 1. A light receiving element having an electrode structure in which a pair of electrodes are opposed to each other on the left and right and a dark current is kept small in a substantially symmetrical direction is provided on a semiconductor substrate, and the two light receiving elements are electrically connected to each other. The connected point serves as an electric signal output terminal, and the two light receiving elements are supplied with bias voltages of opposite polarities and substantially equal in size when viewed from the electric signal output terminal. A circuit that outputs an electric signal from the electric signal output terminal only when light is input to one side is regarded as one unit circuit, and a plurality of the unit circuits are arranged on a semiconductor substrate to be integrated and input. A contour detection system characterized by detecting a contour of an image pattern. 2. The adjacent unit circuits are arranged such that the light receiving regions of the respective light receiving elements overlap each other so that no contour non-detection portion occurs. The described contour detection system. 3. A unit group is formed by arranging a plurality of unit circuits in each of which light-receiving elements are arranged and connected in the longitudinal, lateral, diagonally rightward, and diagonally leftward directions so as to form a unit group, and a plurality of said units are provided. The contour detection system according to claim 1, wherein the arrangement of groups enables contour detection in four directions of an input image pattern. 4. The contour of the color or wavelength of the input image pattern can be detected by forming the response wavelength ranges of one set of the light receiving elements in the unit circuit so as to be different from each other. The contour detection system according to claim 1. 5. The contour detection system according to claim 1, wherein a circuit having a low-pass filter characteristic in time series or space series is connected to the output terminal. 6. The contour detection system according to claim 1, wherein input terminals of a pair of semiconductor light emitting elements are complementarily connected to the output terminals. 7. The contour detection system according to claim 1, wherein the light receiving element has a structure in which rectifying junctions are connected in opposite directions. 8. The contour detection system according to claim 7, wherein the rectifying junction is a Schottky junction.