JPH10326341A - Image sensor - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide the image sensor which can easily perform coordinate transformation for horizontal or vertical compression fast at low cost. SOLUTION: The sensor is equipped with pixels having a rectangular area by arranging pixels each consisting of a photodetection part area (photodiode) 11 and a switch part area (MOSFET 12) so that the photodetection part area and switch part area are adjacent to each other, those pixels are arranged, and the photodetection part areas of the respective pixels are arranged in a specific linear direction adjacently to the photodetection part areas of adjacent pixels. Consequently, each pixel becomes rectangular having a longitudinal/ lateral ratio which is not 1:1; and horizontally compressed transformation is automatically performed by making the pixels laterally long and vertically compressed coordinate transformation is performed by making the pixels longitudinally long. Consequently, the coordinate transformation need not be performed by a CPU and an edge image of a lane can easily be obtained to shorten the total processing time of lane detection.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a so-called area image sensor in which a plurality of pixels are arranged two-dimensionally on a plane, and for example, to a configuration of an image sensor suitable for detecting a lane on a road surface.

[0002]

2. Description of the Related Art Conventionally, a device shown in FIG. 8 is known as a device for detecting a lane on a road surface (a line indicating a boundary between a road and a traveling lane). Explaining the configuration, the output of the video camera 1 for imaging the road is A / D (analog / analog).
The digital signal is A / D-converted by a (digital) converter 3, and the converted digital signal is input to a CPU (central processing unit) 4. 5
Is an image memory for storing image information.

The CPU 4 performs a process as shown in FIG. In other words, first, Sobel
A differential image is obtained by operating a differential operator such as an operator. Next, edges are detected using a method such as comparing the differential value with a predetermined threshold. The lane is detected from the edge by using a method such as comparing the finally detected edge with the lane model.

The above-mentioned video camera 1 uses an image sensor 2. As the image sensor 2,
A so-called area image sensor that has a plurality of pixels, is arranged in a two-dimensional plane, and outputs the intensity of light received by each pixel by scanning is used. A typical image sensor pixel is, for example, "BRYA
N ACKLAND, ALEX DICKINSON, "CAMERA ON A CHIP", I
EEE INTERNATIONAL SOLID-STATE CIRCUITS CONFERENCE
DIGEST OF TECHNICAL PAPERS, VOL.39, P.23, 1996 ".

FIG. 10 is a diagram showing an example of a road image taken by the video camera 1, in which (a) shows a normal image, and (b) compresses the image of (a) in the horizontal direction. FIG. 6 is a diagram showing an example in which coordinate conversion is performed. Here, for example, the video camera 1 is installed behind the rear-view mirror of the vehicle toward the front of the vehicle (for example, the height from the ground is approximately 1.5).
m), and the image is taken around a few tens of meters ahead of the car. In FIG. 10, 6 is a preceding vehicle, 7 is a horizon, and 8 and 9 are lanes indicating boundaries of traveling lanes.

As shown in FIG. 10A, in the road image, in addition to the lanes, for example, obstacles such as the preceding vehicle 6 appear in the image. As the number of edges of these obstacles increases, it becomes more difficult to select a lane from the edges, and the processing for that becomes more complicated. Also, as shown in FIG. 10A, an image of an obstacle such as the preceding vehicle 6 and the horizon 7 includes many horizontal line segments, whereas an image of a lane is either horizontal or vertical. No, it is oblique. Then, FIG.
In the differentiation processing shown in (1), differentiation in the direction orthogonal to the direction of the line produces the most edges. Therefore, if the coordinate conversion is performed so that the image of the lane approaches vertical, a sufficient lane edge image can be obtained only by the horizontal differentiation.

FIG. 10B is a diagram showing an example in which a coordinate transformation of (X, Y) → (X / 2, Y) is performed on the image of FIG. 10A. By performing the conversion for compressing the horizontal direction in this way, a sufficient lane edge image can be obtained only by the horizontal differentiation. In addition, since an edge of an obstacle is hardly obtained by the horizontal differentiation, there is an advantage that a lane can be easily detected from the edge image.

[0008]

SUMMARY OF THE INVENTION In lane detection,
If the coordinate transformation as described above is performed, the edge of the obstacle cannot be obtained much depending on the horizontal differentiation, so that a sufficient lane edge image can be obtained only by the horizontal differentiation, and it is easy to detect the lane from the edge image. There are advantages. However, when the above-described coordinate conversion is performed by the CPU, there is a problem that the processing takes time.

As another method of performing coordinate conversion as described above, conversion using an optical system may be considered.
In this case, a lens system having different vertical and horizontal magnifications is required. For example, such a lens system can be realized by combining a cylinder lens and an ordinary lens. However, such a configuration causes an economic problem of high cost, and is difficult to realize industrially. Further, the above-described lens system has problems such as aberration, difficulty in focusing, and image distortion.

After the original image is A / D converted, the CPU
, The quantization noise generated at the time of A / D conversion is emphasized by the differentiation, so that the obtained differential image contains much noise. Conventionally, a method such as the Sobel operator, in which smoothing and differentiation are performed simultaneously to reduce the influence of noise, has been used, but there is a problem that it takes a long time to perform the processing.

SUMMARY OF THE INVENTION The present invention has been made to solve the problems of the prior art as described above, and a first object of the present invention is to enable high-speed, simple, and inexpensive coordinate transformation for compressing in the horizontal or vertical direction. A second object is to provide an image sensor that can easily perform analog differentiation in the same direction as the scanning direction.

[0012]

Means for Solving the Problems In order to achieve the above object, the present invention is configured as described in the claims. That is, according to the first aspect of the present invention, the light receiving unit includes a light receiving unit area and a switch unit area that permits output of the light receiving unit, and the light receiving unit area and the switch unit area are arranged adjacent to each other. By doing so, a plurality of pixels having a rectangular area are provided, the pixels are arranged two-dimensionally in a plane, and the light receiving area of each pixel is the same as the light receiving area of a pixel adjacent in a predetermined one-dimensional direction. The configuration is such that they are arranged side by side. According to the above configuration, the shape of each pixel is a rectangle having an aspect ratio other than 1: 1. With this configuration, FIG.
As will be described in detail with reference to FIG. 2, if the pixel is made horizontally long, the coordinate conversion in which the horizontal direction is compressed, and if the pixel is made vertically long, the coordinate conversion in which the vertical direction is compressed is automatically performed without the coordinate conversion processing in the CPU. Will be. Therefore CPU
Eliminates the need to perform the above coordinate transformation process,
While suppressing edges such as obstacles, a sufficient lane edge image can be obtained only by horizontal differentiation, and the overall processing time of lane detection can be reduced. Further, if the differentiation in the vertical direction is performed, detection of an obstacle or the like becomes easy. Further, since the pixel shape of the image sensor is merely formed in a rectangular shape, a special lens system or the like is not required, and the image sensor can be realized at low cost. Note that since the above vertical and horizontal directions correspond to the horizontal / vertical directions when installed in a video camera, one type of actual image sensor may be used, and when installed in a video camera, the longitudinal direction is set to the horizontal or vertical direction. Thus, it can be arbitrarily selected.

Further, in the invention according to claim 3,
When the signal of each pixel of the image sensor is read, the difference between the intensity of the light read at the current clock and the intensity of the light read one clock before is used as the output of the image sensor. With the above configuration, analog differentiation can be performed in the same direction as the scanning direction with the minimum number of additional circuits, and A / D conversion can be performed after the differentiation. -It is possible to omit an operation requiring a long processing time such as an operator.

The inventions according to claims 4 and 5 show specific circuit configurations for realizing the configuration of claim 3. As the conversion circuit according to the fifth aspect, for example, a buffer circuit, an integration circuit, a peak detection circuit, or the like can be used.

[0015] The invention described in claim 6 is the first invention.
Alternatively, the image sensor according to claim 2 is combined with any one of claims 3 to 5, and differential processing can be performed in horizontal scanning or vertical scanning, so that noise is reduced in a short time. An edge image of a lane or an obstacle can be obtained, and the processing time for detection can be reduced.

[0016]

According to the first and second aspects of the present invention, the coordinate conversion is performed at the same time as the image data is read into the image memory. As a result, it becomes easier to distinguish between the lane and the obstacle. The overall time can be reduced. Further, a sufficient lane edge image can be obtained only by horizontal differentiation, and time-consuming processing such as omnidirectional differentiation can be omitted. In addition, since the shape of the pixel of the image sensor is simply formed in a rectangular shape, there is obtained an effect that a special lens system or the like is not required and the image sensor can be realized at low cost. Similarly, an edge image of an obstacle such as a preceding vehicle can be obtained only by differentiation in the vertical direction.

According to the third to fifth aspects of the present invention, the differentiation in the same direction as the scanning direction can be realized with a minimum number of additional circuits.
D conversion can be performed, and quantization noise can be prevented from being emphasized by differentiation. Therefore, an effect is obtained that a differential image with less noise is obtained.

In the invention according to claim 6,
By combining the first and second aspects and the third to fifth aspects, differentiation processing can be performed in horizontal scanning, so that a lane edge image with less noise can be obtained in a short time, and the processing time of lane detection can be reduced. The effect of shortening can be obtained. Similarly, since the differentiation processing can be performed in the vertical scanning, an edge image of an obstacle such as a preceding vehicle can be easily obtained.

[0019]

Embodiments of the present invention will be described below. FIGS. 1A and 1B are diagrams showing an image sensor according to a first embodiment of the present invention, wherein FIG. 1A is an equivalent circuit diagram, and FIG. 1B is a plan pattern diagram of an element. This example shows a case where the present invention is applied to a MOS image sensor. The equivalent circuit shown in (a) is the same as that of a conventional MOS image sensor, and each photodiode 11 is connected to a vertical signal line 13 via a MOSFET 12. Each MOSFET1
The two gates are connected to the vertical selection line 14. In this case, one photodiode 11 and one MOSFET
12 form a pixel. The photodiode 11 corresponds to a light receiving area, and the MOSFET 12 corresponds to a switch area, and both constitute one pixel.

Further, as shown in FIG. 2B, each pixel is arranged so that the photodiode 11 and the MOSFET 12 are arranged adjacent to each other in the horizontal direction. Comrade, MOSFE
The regions T12 are arranged so that they are adjacent to each other. That is, in one pixel, the photodiode 11 and the MOSFET 12 are arranged in the horizontal direction, and only the photodiode and the MOSFET are arranged in the vertical direction. Therefore, the shape of each pixel is a rectangle (for example, the aspect ratio is 1: 2) in which the horizontal direction (X) is longer than the vertical direction (Y). In FIG. 1B,
The photodiode 11 (light receiving area) itself has a rectangular shape that is long in the horizontal direction. However, the present invention is not limited to this. It is only necessary that the entire shape of one pixel including the photodiode 11 and the MOSFET 12 be rectangular. FIG. 1 shows only four pixels, but does not show that many pixels exist.

Although FIG. 1 illustrates a case where the pixels are horizontally long, the pixels may be vertically long. However, since the above horizontal and vertical directions correspond to the horizontal / vertical directions when installed in a video camera, one type of actual image sensor may be used, and when installed in a video camera, the longitudinal direction is set to the horizontal or vertical direction. Thus, it can be arbitrarily selected.

Next, the operation will be described. First, assuming that a certain vertical selection line 14 is selected by vertical scanning, the light passes through the photocharge MOSFET 12 accumulated in the photodiode 11 and flows out to each vertical signal line 13. Next, by the horizontal scanning, the photocharge flowing into the vertical signal line flows to the output through a horizontal scanning MOSFET (not shown). This output passes through the A / D converter 3 in FIG.
4 are sequentially read into the image memory 5. Generally, since the image memory is vertically and horizontally symmetric, the image of the read image has an aspect ratio of 1: 1. On the other hand, the pixels of the image sensor according to the present embodiment have an aspect ratio of 1: 2. Therefore, just by reading, the horizontal direction is 1 /
Can be compressed to 2.

FIG. 2 is a diagram showing the above state, and FIG.
Is a conventional image sensor (aspect ratio is 1: 1),
(B) shows the case of the image sensor of FIG. 1 (the aspect ratio is 1: 2). The left figure shows the image on the image sensor, and the right figure shows the image on the image memory. As can be seen from FIG. 2, in the conventional image sensor shown in FIG. 2A, the image on the image sensor and the image on the image memory are the same. However, in the present invention shown in (b), since each pixel of the image sensor is horizontally long, if it is stored in a vertically and horizontally symmetrical image memory, the horizontal (horizontal) direction as shown in the right figure of (b) is obtained. Direction) is a compressed image. That is, at the time when the data is simply read from the image sensor and stored in the image memory, the coordinate conversion in which the horizontal direction is compressed is performed.

As described above, according to the present invention, the horizontal direction can be compressed without a complicated lens system or coordinate conversion processing in the CPU. As a result, the image of the lane appearing in the oblique direction can be made closer to the vertical, so that the lane can be easily distinguished from an obstacle or the like, thereby shortening the entire processing time of the lane detection. Furthermore, since the lane image approaches vertical, a sufficient lane edge image can be obtained only by the horizontal differentiation, and the omnidirectional differentiation that takes much processing time can be omitted, and the processing direction can be fixed because the differentiation direction can be fixed. It can be further shortened.

The above example shows a configuration effective for lane detection. However, if the scanning direction is changed to the vertical direction with the same configuration as the above, a nearly vertical lane is not detected. (Such as obstacles) can be easily detected.
Therefore, an edge image such as an obstacle with little noise can be obtained in a short time. That is, in the present invention, it is possible to easily detect either a line close to the vertical direction or a line close to the horizontal direction simply by changing the scanning direction.

Further, in the example of FIG. 1, the case where the aspect ratio of the pixel is set to the horizontal length of 1: 2 is shown. However, as can be understood from the above description, the vertical length is effective. In the above example, M
Although the case where the present invention is applied to the OS type image sensor has been illustrated, the present invention can be similarly applied to other image sensors, for example, a CCD or the like.

FIG. 3 is a block diagram showing a second embodiment of the present invention.

In FIG. 3, reference numeral 20 denotes an image sensor,
1 is a delay circuit, 22 is a differential amplifier. The image sensor 20 may have the structure of the present invention shown in FIG. 1 or an image sensor having a conventional structure. In FIG. 3, the output of the image sensor 20 is input to a delay circuit 21 to generate a signal delayed by one scanning clock, and the difference between the signal and the signal read out at the current clock is used as the differential amplifier 2.
2 and the output of the differential amplifier 22 is sent to the A / D converter 3 in FIG.

Next, the operation will be described. When a certain pixel is accessed when the image sensor is scanned in the horizontal direction, the output of the differential amplifier 22 outputs the output one clock before, that is, the output of the pixel adjacent to the pixel in the horizontal direction, and the current output, that is, the output of the pixel. The difference from the output is output. This value corresponds to performing analog differentiation in the same direction as the scanning direction. Therefore, the image sensor shown in FIG. 3 outputs an analog differential value in the same direction as the scanning direction, and the value is A / D converted by the A / D converter 3 in FIG. Become.

As described above, in the embodiment shown in FIG. 3, since the differentiation is performed before the A / D conversion, a differential image with little noise is obtained without emphasizing the quantization noise. For example, it is assumed that the output of a certain pixel and the adjacent pixel is 0.8 and 0.4, respectively. When analog differentiation is performed first as in this embodiment, the result is 0.4, and when A / D conversion by rounding is performed, the result is 0. On the other hand, as in the conventional example shown in FIG.
, The respective outputs become 1.0, and the result of the differentiation becomes 1. Here, it is assumed that the DC level slightly fluctuates, and the output of each pixel becomes 0.9 and 0.5. When the A / D conversion is performed after the differentiation, the result becomes 0, indicating that there is no influence of the DC fluctuation. On the other hand, when the A / D conversion is performed first, the result of the differentiation becomes 0, and it can be seen that the result of the differentiation greatly moves due to a weak fluctuation of the direct current.

As described above, in the present embodiment,
Since a differential image with less noise is obtained, smoothing means such as a Sobel operator is not required, and the time required for differentiation can be reduced. Note that it is desirable to arrange the delay circuit 21 and the differential amplifier 22 as close to the image sensor 20 as possible or to be built in the image sensor 20 in order to avoid mixing of noise due to induction or the like.

The image sensor 20 shown in FIG.
The use of the image sensor shown in FIG. 1 according to the first embodiment of the present invention achieves a still greater effect. That is, when the image sensor of FIG. 1 is used, the image of the lane can be made closer to the vertical, so that only the horizontal differentiation is sufficient. Therefore, if the configuration is as shown in FIG. 3 and scanning is performed in the horizontal direction, coordinate conversion and differentiation are performed simultaneously with reading into the image memory, and a lane edge image with little noise can be obtained in a short time.

The above example shows a configuration effective for lane detection. However, if the scanning direction is changed to the vertical direction with the same configuration as above, a lane close to vertical is not detected, so that noise is reduced in a short time. An edge image of an obstacle or the like with a small number is obtained.

FIG. 4 is a block diagram showing a third embodiment of the present invention. In the example shown in FIG. 3, only a differential image can be output. However, for example, when the electronic shutter is controlled using the brightness of the entire image, and the control is performed such that the brightness is within a predetermined range by shortening the shutter time when the image is too bright, the original image is not differentiated. You need an image. The example shown in FIG. 4 is such that the differential image and the original image can be switched and output. FIG.
In the figure, reference numeral 23 denotes a selection switch, and the same reference numerals as those in FIG. 3 denote the same parts. In the circuit of FIG. 4, the output of the delay circuit 21 and the fixed potential V ₀ can be switched as the input of the differential amplifier 22. If you want to output the original image, it is given a fixed potential V ₀ which switches the changeover switch 23 to the input of the differential amplifier 22,
The output of the differential amplifier 22 is the output of the pixel and the fixed potential V
The difference from ₀ , that is, the original image.

Next, FIG. 5 is a block diagram showing a fourth embodiment of the present invention. This example shows another embodiment in which an original image can be switched and output. In FIG. 5, reference numeral 24 denotes a dark cell (a cell not exposed to light) provided in the image sensor 20, and the same reference numerals as those in FIG.

In FIG. 5, in order to prepare a dark cell 24 which is not exposed to light in the image sensor 20 and to read an original image of a certain pixel, first read the dark cell 24 and then scan so as to read the pixel. I do. In this way, all outputs one clock before output from the delay circuit 21 are dark cells 2
4 (0), the output of the differential amplifier 22 becomes the output of the pixel, and an original image is obtained. When it is desired to output a differential image, the same process as in FIG. 3 may be performed without performing the process of reading the dark cell 24. When the brightness of the entire image is required, the differential image can be obtained by integrating the differential image twice. However, in this case, the output of a dark cell serving as a DC reference is required.

FIG. 6 is a circuit diagram showing a fifth embodiment of the present invention, and shows an actual configuration example of the circuit shown in FIG. In FIG. 6, the original output of the image sensor 20 is connected to two capacitances C1 and C2 having the same characteristics such as capacitance via switches S1 and S2, respectively, and the capacitances C1 and C2 are respectively connected to the switches S3 and S3. It is connected to the + input and-input of the differential amplifier 25 via S4. The output of the differential amplifier 25 becomes the output of the image sensor and is sent to the CPU 4 via the A / D converter 3 in FIG.

Next, the operation will be described. Image sensor 2
When one pixel of 0 is selected, a signal charge proportional to the intensity of the input light of that pixel appears as an original output. If the switches S1 and S2 are turned on at the same time, the switches S3 and S4 are turned off, and the capacitances C1 and C2 have the same characteristics, equal charges proportional to the original output charges are respectively stored in C1 and C2. It is stored. Next, the switch S5 is turned on to reset the electric charge of the parasitic capacitance C3 at the + input of the differential amplifier 25. Subsequently, the switch S5 is turned off, and the switch S3 is turned off.
Turn on. Then, a charge proportional to the charge stored in the capacitance C1 is transferred to the + input of the differential amplifier 25. At this time, the previous clock (1
Since the data before the clock is stored, the differential amplifier 25 outputs the difference between the data of the current clock and the data of the previous clock.

Next, before selecting the next pixel of the image sensor 20, the switch S6 is turned on and the differential amplifier 25 is turned on.
Then, the charge of the parasitic capacitance C4 at the-input is reset, and then the switch S6 is turned off and the switch S4 is turned on. As a result, a charge equal to the charge stored at the + input of the differential amplifier is transferred from the capacitance C2 to the-input of the differential amplifier, and the data of the current clock is stored. The signal transfer from the differential amplifier 25 to the A / D converter 3 is linked to the on / off of the switch, and the differential amplifier 25 outputs the difference between the current clock data and the previous clock data. Perform when you are. In this circuit, it is necessary to reset before storing signals in the capacitances C1 and C2. The switches S7 and S8 are the reset switches described above.

Next, FIG. 7 is a circuit diagram showing a sixth embodiment of the present invention, and shows another example of the actual configuration of the circuit shown in FIG. In the case of FIG. 6, the capacitances C1 and C
2, 4 data transfer switches (S1, S2, S3, S
4) and four reset switches (S5, S6, S
7, S8) is required. In the example of FIG. 7, the above components are reduced. In FIG. 7, the image sensor 20
Is connected to the input of the buffer circuit 26, and the output of the buffer circuit 26 is connected to the + input and the-input of the differential amplifier 25 via two switches S3 and S4, respectively. The original output of the image sensor 20 is integrated by the input parasitic capacitance C5 of the buffer circuit 26, and a voltage proportional to the integrated value is output from the buffer circuit 26.

First, when the switch S3 is turned on, the voltage output of the buffer circuit 26 is applied to the + input of the differential amplifier 25. At this time, since the output impedance of the buffer circuit 26 is very low, the voltage output of the buffer circuit 26 is stored in the parasitic capacitance C3 regardless of the residual charge of the parasitic capacitance C3. At this time, since the data of the previous clock (one clock before) is stored in the minus input (parasitic capacitance C4) of the differential amplifier 25, the differential amplifier 25 calculates the difference between the data of the current clock and the data of the previous clock. Is output. Next, S
When 3 is turned off and S4 is turned on, data of the current clock is stored in the minus input (parasitic capacitance C4) of the differential amplifier 25.

Next, before selecting the next pixel of the image sensor 20, the switch S9 is turned on and the buffer circuit 2 is turned on.
6, the charge of the input parasitic capacitance C5 is reset, and the switch S9 is subsequently turned off. Thus, the output of the buffer circuit 26 becomes new clock data. The signal transfer from the differential amplifier 25 to the A / D converter 3 is linked to the on / off of the switch, and the differential amplifier 25 outputs the difference between the current clock data and the previous clock data. Perform when you are. As described above, since the buffer circuit 26 outputs a voltage signal and has low output impedance, it is not necessary to reset the charges of the parasitic capacitances C3 and C4 of each input of the differential amplifier 25. Therefore, the number of switches can be reduced, and the control can be simplified as compared with FIG.

In FIG. 7, the buffer circuit 26
Although the original output is detected and converted into a voltage signal using the parasitic capacitance of the above, the same operation can be performed by using an integration circuit or a peak detection circuit instead of the buffer circuit 26. In short, any circuit that converts the output of the pixel of the image sensor 20 into a voltage signal and has a low output impedance may be used.

In the embodiment shown in FIGS. 3 to 7,
Scanning in the horizontal direction has been described as an example, but scanning in the vertical direction does not detect lanes that are close to vertical, so that edge images such as obstacles with little noise (many horizontal line segments) can be obtained in a short time. Can be

For example, when it is desired to determine whether the object ahead is a vehicle, since the upper surface of the bumper or trunk of the vehicle is a line extending relatively in the horizontal direction, it is necessary to detect a vertical edge component from the detected image. However, the shape varies depending on the vehicle, and there are irregularities. In such a case, each pixel is configured vertically (photodiode 11
And the FOSFETs 12 are arranged in the vertical direction in each pixel, and each pixel is vertically compressed by disposing the photodiodes 11 and the FOSFETs 12 adjacent to each other in the horizontal direction. Therefore, the edge component in the horizontal direction is emphasized, so that the bumper and the upper surface of the trunk can be detected with higher accuracy, and the accuracy of vehicle determination is improved.

In the above description, the present invention is applied to the case where the present invention is applied to an image sensor for detecting an obstacle in the lane or the forward direction. Without this, compression in the horizontal direction (when the pixels are horizontally long) or compression in the vertical direction (when the pixels are vertically long) can be performed. Therefore, it can be applied to applications requiring such characteristics. In addition, since the image of a line appearing in an oblique direction can be made closer to vertical or horizontal, an edge image of a line close to the vertical direction or a line close to the horizontal direction can be obtained only by horizontal differentiation or vertical differentiation. In addition, the omni-directional differentiation that requires a lot of processing time can be omitted, and the direction of the differentiation can be fixed, so that the processing time can be further reduced.

[Brief description of the drawings]

FIGS. 1A and 1B are diagrams showing an image sensor according to a first embodiment of the present invention, wherein FIG. 1A is an equivalent circuit diagram, and FIG.

2A and 2B are diagrams illustrating image states of a conventional image sensor and an image sensor of the present invention, wherein FIG. 2A illustrates a conventional image sensor (aspect ratio is 1: 1), and FIG. 2B illustrates an image sensor of FIG. The case where the aspect ratio is 1: 2) is shown. The left figure shows the image on the image sensor, and the right figure shows the image on the image memory.

FIG. 3 is a block diagram showing a second embodiment of the present invention.

FIG. 4 is a block diagram showing a third embodiment of the present invention.

FIG. 5 is a block diagram showing a fourth embodiment of the present invention.

FIG. 6 is a block diagram showing a fifth embodiment of the present invention.

FIG. 7 is a block diagram showing a sixth embodiment of the present invention.

FIG. 8 is a block diagram of an example of a conventional lane detecting device.

FIG. 9 is a flowchart showing processing calculations in the CPU 4 of FIG. 8;

FIGS. 10A and 10B are diagrams illustrating an example of a road image captured by the video camera 1, in which FIG. 10A illustrates a normal image, and FIG. 10B illustrates a coordinate transformation for compressing the image of FIG. The figure which shows the example which performed.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 ... Video camera 2 ... Image sensor 3 ... A / D converter 4 ... CPU 5 ... Image memory 6 ... Previous vehicle 7 ... Horizon 8, 9 ... Lane indicating the end of the traveling lane 11 ... Photodiode 12 ... MO
SFET 13 Vertical signal line 14 Vertical selection line 20 Image sensor 21 Delay circuit 22 Differential amplifier 23 Switching switch 24 Dark cell 25 Differential amplifier 26 Buffer circuit S1 to S4 Switching switches S5 to S9 ... Reset switches C1 and C2 ... Capacitance C3 and C5 ... Parasitic capacitance

Claims (6)

[Claims] 1. A rectangular area comprising a light receiving section area and a switch section area permitting output from the light receiving section, wherein the light receiving section area and the switch section area are arranged so as to be adjacent to each other. Are arranged in a two-dimensional plane, and the light receiving area of each pixel is adjacent to the light receiving area of a pixel adjacent in a predetermined one-dimensional direction. An image sensor, comprising: an image sensor; 2. The image sensor according to claim 1, further comprising means for reading data of each pixel and storing the data in an image memory.
An image sensor, wherein each dot of the image memory and each pixel of the image sensor have a one-to-one correspondence. 3. It has a plurality of pixels, and the pixels are on a plane 2
An image sensor that is arranged in a dimensional manner and reads out the intensity of light received by each pixel by scanning. The image sensor reads the difference between the intensity of light read at the current clock and the intensity of light read one clock before. An image sensor, wherein the image sensor is configured to output the output of the sensor. 4. An original output of the image sensor is connected to two capacitances having the same characteristic via two switches, respectively, and each capacitance is further connected to two inputs of a differential amplifier via separate switches. Are connected to each other,
The original output accumulated in the two capacitances is given to the two inputs of the differential amplifier with a difference of one clock by the switching control of each switch, and the output of the differential amplifier is set to the output of the image sensor. The image sensor according to claim 3, wherein the image sensor is configured to perform the operation. 5. An image sensor according to claim 1, further comprising a conversion circuit for converting an original output of the image sensor into a voltage signal and having a low output impedance. Connected to each of two inputs, and by switching control of each switch, a voltage signal corresponding to the original output is given to two inputs of the differential amplifier with a difference of one clock, and the output of the differential amplifier is imaged. 4. The image sensor according to claim 3, wherein the image sensor is configured to output the output of the sensor. 6. An image sensor according to claim 3, wherein the image sensor according to claim 1 or 2 is used as the image sensor for transmitting the original output. .