JPH11337493A - Detecting system for position of abnormal point by image processing - Google Patents

Abstract

PROBLEM TO BE SOLVED: To obtain a detecting system in which the low resolution of an infrared camera is supplemented with a visible-region ordinary camera and by which an abnormality is analyzed minutely, by a method wherein an image which is photographed by the infrared camera and an image which is photographed by the ordinary camera are converted in a prescribed manner so as to be compared or composed. SOLUTION: A photographing part which acquires the image of a road surface 10, and a processing part which processes the image so as to find the position of an abnormal point, constitute an abnormal-point detecting system which is applied for the diagnosis of a leak of water on the road surface 10 and which relies on an image processing operation. Then, e.g. in the photographing part, the road surface 10 or the like is photographed by an infrared camera 30 in which the center of respective photographing ranges is set so as to agree approximately and by a visible-region digital camera 40, and images are recorded as pieces of digitized information. In the processing part, on the basis of an image which is photographed in a prescribed reference state and on the basis of an image which is used to detect the position of the abnormal point, a correction conversion parameter by which an oblique image with reference to the road surface 10 is converted into an orthogonal projection image is decided. The respective camera images are converted by the parameter, the images are composed, insert-composed or the like, and the abnormal point is specified.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an abnormal point position detection system by image processing for detecting an abnormal point position by processing an image by infrared rays. The present invention relates to a system and method for detecting an abnormal point position by image processing for easily specifying the position of a detected abnormal point.

[0002]

2. Description of the Related Art Conventionally, as a means for diagnosing abnormalities on a road surface, for example, water leakage from a water pipe buried underground, cracks or peeling of a pavement, an infrared camera (IR camera) is used to photograph a predetermined position on the road surface. A possible reference object is arranged, an image of the road surface is taken by an IR camera so that the reference object is included in the screen, and the position of the abnormal point on the road surface which is grasped from the temperature distribution of the road surface detected by the image is determined by the reference object. There is known a method of making a diagnosis based on the position of an object.

[0003] Also, there is known software for creating a panoramic photograph by connecting images based on image data of points included in the overlapping portion, using digital images photographed with overlapping photographing ranges.

[0004]

In the conventional system for detecting an abnormal point using an IR camera, since the number of pixels of the detection element of the IR camera is tens of thousands, the resolution of the image by the IR camera is low. Since the position of the reference point on the image and the state of the temperature distribution on the inspection object are unclear, it has been difficult to specify and analyze the abnormal part and the abnormal state in detail.

[0005] In order to compensate for the low resolution of the image by the IR camera, the same part as that of the IR camera may be photographed by a normal camera. However, each camera is independent, and the photographing magnification and the photographing magnification are different. Since the state of the image distortion due to the range and tilt photographing is different, it is only for reference, and the state of the inspection object cannot be analyzed in detail. In addition, even if an attempt is made to panorama an image in order to grasp the situation of the entire inspected object to be diagnosed, the positions of points having known outlines and relative positions of shapes serving as a reference for panorama formation are not clear as described above. Therefore, the above-mentioned processing was difficult.

Accordingly, an object of the present invention is to compensate for the low resolution of an image obtained by an IR camera by using an image of a normal camera, and to analyze in detail the abnormal state of the inspection object and its position. An object of the present invention is to provide an abnormal point position detection system by image processing.

[0007]

According to a first aspect of the present invention, there is provided an infrared camera which is installed at a predetermined relative position, and which captures an imaging range including a range common to both with a tilt angle. A shooting unit consisting of a normal camera,
A processing unit that processes the image of the infrared camera and the image of a normal camera, wherein the processing unit is configured to capture a plurality of reference points whose relative positions within the common shooting range are known. 1 based on the position of the reference point in the first image captured by the normal camera, between the first image and the first image of the normal camera. A first conversion parameter determining unit for obtaining a first correction conversion parameter for correcting and converting a first image of the normal camera into an orthographic projection image; and a distortion of an image caused by the tilt photographing. A second conversion for obtaining a second correction conversion parameter for correcting and converting the first image shot by the infrared camera into an orthographic projection image based on an image based on shooting by the normal camera; A calibration unit comprising a parameter determination unit, and a second image of the infrared camera captured so as to include a plurality of points whose relative positions are known in the common shooting range of the infrared camera and the normal camera. And a second image of the normal camera, based on the position of the relative position known point in the second image photographed by the normal camera, having an image distortion generated by the tilt photographing. A third conversion parameter determining unit for obtaining a third correction conversion parameter for correcting and converting the second image of the normal camera into an orthographic projection image; the second correction conversion parameter; and the third correction A fourth conversion parameter determining unit for obtaining a fourth correction conversion parameter for correcting and converting the second image of the infrared camera into an orthographic projection image based on a difference from the conversion parameter; It is abnormal point position detection system according to the image processing comprising the composed image correcting unit.

[0008] An image taken by a normal camera used for obtaining the second correction conversion parameter may be an image having a distortion of an image generated by the tilt shooting or a positive image converted by the first correction conversion parameter. It can be a projection image. The image correction unit corrects and converts the second image of the infrared camera with the fourth correction conversion parameter, and / or converts the second image of the normal camera with the third correction conversion parameter. It is preferable that the image processing apparatus further includes an image conversion unit that performs the correction conversion and generates an orthographic projection image.

Since the image of the IR camera for diagnosing the abnormality is converted into an orthographic projection image, the position of the abnormal point on the object is easily specified. Furthermore, if the image of a normal camera is also converted into an orthographic projection image and the two images are compared, the position of the abnormal point on the object can be more easily specified. Further, by determining the ratio between the value on the orthographic projection image of the dimension between the known points and the actual dimension, the actual coordinate value of the abnormal point,
It is also possible to know the size (length, area, etc.) of the abnormal part.

In any one of the above inventions, an orthographic projection image of a second image of a normal camera converted by using the third correction conversion parameter is obtained by using the fourth correction conversion parameter. An image synthesizing unit may be further provided for creating an image in which the converted orthographic image of the second image of the infrared camera is replaced with a corresponding part of the orthographic image of the second image of the normal camera. . by this,
The position of the abnormal point is specified more easily.

In the case of diagnosing a road surface abnormality by the above system, a structure provided on or near the road surface such as a center line or a curb is used as a reference point.
This is preferable because it is not necessary to set a special reference point.

[0012]

DESCRIPTION OF THE PREFERRED EMBODIMENTS As an embodiment of the present invention, an abnormal point position detecting system using image processing for detecting the position of an abnormal point on a road surface will be described. However, such embodiments do not limit the technical scope of the present invention. The abnormal point position detection system using the image processing of the present embodiment is a photographing unit that photographs the road surface of the observation target road and acquires the image, and the acquired image information is obtained in order to obtain the position of the abnormal point on the road surface. And a processing unit for processing the image.

FIG. 1 is a diagram showing an outline of a photographing state of a road surface by the photographing section. Moving means 2 for moving on road surface 10
0 is provided with an IR camera 30 and a digital camera 40. The position of each camera and the direction of the optical axis of the IR camera 30 and the digital camera 40 are set so that the centers of the respective photographing ranges substantially coincide with each other. The optical axis of the digital camera 40 faces downward at an unknown angle α1 with respect to the horizontal when the road surface is horizontal. The optical axis of the IR camera 30 faces downward at an unknown angle α2 with respect to the horizontal when the road surface is horizontal. Usually, α1 and α2 do not match.

An image from the IR camera 30 and an image from the digital camera 40 are recorded as digitized image information. However, the image information obtained by any camera may be obtained by digitizing an image photographed on a film by a film scanner or the like. In this system,
First, in a predetermined reference state, the image capturing unit captures a reference point of a known position or size, and performs calibration for obtaining first and second correction conversion parameters for converting images of each camera of the image capturing unit into orthographic projection images. Perform an action.

Then, an abnormal point position is detected. In the abnormal point position detection, a road surface having a reference point whose position or dimension is known, which cannot be always specified in the image of the IR camera 30 but can be specified in the image of the digital camera 40, is taken in an arbitrary state, and the image of the digital camera 40 is taken. A third correction conversion parameter to be converted to an orthographic projection image is obtained, and a correction based on a difference between the first correction conversion parameter and the third correction conversion parameter is added to the second correction conversion parameter. 4 is obtained.

Further, the image photographed in the arbitrary state is corrected and converted by the third and / or fourth correction conversion parameters to obtain a desired image. First to fourth conversion parameter determination units for obtaining the first, second, third, and fourth correction conversion parameters; and correction correction conversion of an image captured by each camera using the correction conversion parameters. The image conversion unit for obtaining an image is performed by a processing unit including a personal computer or the like. The first to fourth conversion parameter determination units and the image conversion unit will be described later in detail.

First, the calibration will be described. I at a predetermined relative position on the road surface 10
A plurality of reference objects whose positions can be clearly identified also from the image of the R camera 30 are identified by the IR camera 30 and the digital camera 4.
Shoot at 0. As the reference object, for example, a disk or a cross-shaped plate that forms an image on a pixel of the IR camera 30 in a size covering a plurality of pixels can be used. By processing the image of the IR camera 30 to determine the center position of the circular or cross shape, the position of the reference object can be determined with high accuracy even for the image of the IR camera 30 with low resolution.

Since the IR camera 30 has a high temperature resolution, it can detect the above-mentioned reference object which is usually arranged on the road surface. However, if necessary, it can be easily detected by heating or cooling beforehand. Keep it. Also,
The reference object may be an existing structure having a known size or relative position installed on or near the road surface, for example, a road anchor indicating the center of the road width.

In the following description, a case will be described in which the reference object is arranged at each vertex of a rectangle having a predetermined size on a substantially horizontal road surface. FIG. 2 is an explanatory diagram illustrating a state of an image obtained by photographing the reference object with the digital camera 40. The photographing of the road surface by the digital camera 40 is performed with the tilt angle = α1. As a result, the captured image becomes an image distorted by the tilt angle.

That is, in FIG. 2, reference numeral 41 denotes the size (range) of an image 42 photographed by the digital camera 40. The plane shape of the range of the actual road surface photographed in the image 42 is the range indicated by the one-dot chain line 43. The image 42 includes images 44a, 44b, 44c and 44d of the reference object arranged at the vertices of the rectangle.
Actually, reference object images 44a to 44 which are vertices of a rectangle
d is the position of the vertex of the trapezoid 45 shown by the broken line due to the image distortion due to the tilt photographing. The image photographed by the IR camera 30 is also distorted similarly to the image photographed by the digital camera 40 due to the tilt angle.

FIG. 3 shows a processing section 10 constituted by a personal computer.
FIG. 3 is a diagram showing a configuration of a zero. The processing unit 100 includes a correction conversion parameter determination unit 110 and an image conversion unit 120. The correction conversion parameter determination unit 110 includes a first correction conversion parameter determination unit 110A, a second correction conversion parameter determination unit 110B, and a third correction conversion parameter determination unit 11.
0C, and a fourth correction conversion parameter determination unit 110D.

First correction conversion parameter determination section 110A
And the second correction conversion parameter determination unit 110B constitute a calibration unit. Third correction conversion parameter determination unit 110C and fourth correction conversion parameter determination unit 11
OD and the image conversion unit 120 constitute an image correction unit. The processing unit 100 includes a memory 130 that stores the correction conversion parameters obtained by the correction conversion parameter determination unit 110 and the like.

Here, the image information of the image 42 from the digital camera 40 is converted into a first correction conversion parameter determining unit 11.
0A, and a first correction conversion parameter for correcting the tilting distortion is obtained by software. The principle of the correction of the tilt photographing distortion will be described with reference to FIG. FIG. 4 is an image obtained by photographing a reference object disposed at the vertex of the rectangle. The reference object is an image located at the vertices 44a to 44d of the trapezoid due to the tilting distortion. 2, the same components as those in FIG. 2 are denoted by the same reference numerals, and the description will be simplified.

As described above, the actual shape connecting the points of the reference object is a rectangle having a known size (for example, A * Bcm). Therefore, the points 44 corresponding to the vertices 44a to 44d of the trapezoid 44
a1 to 44d1 are sides 44ab passing through the points 44a and 44b
A first correction conversion parameter for converting 44ab: 44bc1 = A: B on the vertical line is calculated. Here, 44a1 and 44b1 after the correction conversion are 44a and 44a1, respectively.
44b. A rectangle 44a1-44b1-44c1-44d1 connecting the points converted as described above.
Is a reduced orthographic figure having a shape similar to a rectangle connecting the points of the reference object and having a shape connecting the reference points.

Note that the first correction conversion parameter is:
44ab: 44bc1 = A: B * M. Here, M is an arbitrary constant indicating a scale applied only in one direction, here, the longitudinal direction of the center line. That is, the first correction conversion parameter is a correction conversion parameter for correcting a tilting distortion of an image of the digital camera 40 and converting the image into an orthographic projection image.

Next, the image obtained by the IR camera 30 is read into the second correction conversion parameter determination unit 110B of the processing unit 100, and the position of the reference point of the image obtained by the IR camera 30 is set to the position of the reference point of the image obtained by the digital camera 40. A second correction conversion parameter for converting an image obtained by the IR camera so as to match is determined. At this time, the digital camera 4
The image before the correction conversion or the image after the correction conversion can be used as the image based on 0. The image after the correction conversion is obtained by the image conversion unit 120 correcting and converting the image of the digital camera with a first correction conversion parameter. The second correction conversion parameter obtained using the image before the correction conversion is a correction conversion parameter for converting the image of the IR camera 30 into an image having the same image distortion as the image of the digital camera 40. If this image is further converted using the first correction conversion parameter, an orthographic projection image having the same scale as the image of the digital camera 40 can be obtained.

The second image obtained by using the image after the correction conversion
Are the conversion parameters for converting the image of the IR camera 30 into an orthographic projection image of the same scale as the image of the digital camera 40. If the IR camera 30 and the digital camera 40 are relatively rotated around the optical axis, as a result, the IR camera 3
Even if the image of 0 is rotated with respect to the image of the digital camera 40, the rotation relationship is resolved by determining the second correction conversion parameter as described above.

The first correction conversion parameter and the second correction conversion parameter are stored in the memory of the personal computer. The IR camera 30 and the digital camera 40
Are installed close to each other and the difference between the two image distortions is small, the first correction conversion parameter obtained based on the image of the digital camera may be used as the second correction conversion parameter as it is. The same applies to the case where the optical axis of the IR camera 30 and the optical axis of the digital camera 40 are installed so as to be parallel.

Next, detection of an abnormal point position will be described. In the detection of the abnormal point position, the IR camera 30 and the digital camera 40 provided in the moving means 20 capture an image of a target whose presence or absence of abnormality is to be detected. In this case, the IR camera 30
The reference object whose position or size can be specified by the digital camera 40 is not necessarily required. The digital camera 40 is a visible image, and a digital camera 40 capable of obtaining a high-definition image exceeding 1 million pixels is commercially available.
Then, the reference object that can be captured can be easily selected.

Hereinafter, as an example, a case will be described in which an abnormal point position on a road surface including the center line of the road surface as the reference object is detected. Note that it is desirable that the IR camera 30 and the digital camera 40 be photographed at the same time in conjunction with each other, especially when photographing while the moving means 20 is continuously moved. Transportation means 20
This is to eliminate the influence of the fact that the overall posture differs between when photographing with the IR camera 30 and when photographing with the digital camera 40 due to the difference in road surface conditions depending on the location.

FIG. 5A is a photographed image of the road surface including the center line by the digital camera 40. The center line image 50 has a trapezoidal shape due to the image distortion caused by the tilt photographing. By the way, the center line is known to be, for example, a rectangle having a line width * length = 10 cm * 2000 cm. Therefore, the third correction conversion parameter for converting the trapezoidally distorted center line image 50 into a rectangular image similar to the actual center line shape in the same manner as performed in the calibration is calculated by the correction conversion parameter determination unit 110C. Ask.

This third correction conversion parameter generally does not match the first correction conversion parameter. The state in which the photographing for calibration is performed is, for example, a tilt angle, a photographing distance, a posture of the reference object around the optical axis of the camera with respect to the reference object, and a photographing direction to the reference object is different due to unevenness of the road surface. Because. 5B in FIG.
(A) is an image of a center line taken from a different photographing direction, and the trapezoid has a distorted shape. The third correction conversion parameter also corrects a difference in image distortion due to a difference in the photographing direction.

Then, the third correction conversion parameter determination section 110C calculates the first correction conversion parameter and the third correction conversion parameter
, And outputs the difference to the fourth correction conversion parameter determination unit 110D and the image conversion unit 120. Then, the fourth correction conversion parameter determination unit 110
D adds a correction based on the difference from the third correction conversion parameter determination unit 110C to the second conversion parameter to obtain a fourth correction conversion parameter,
20. The fourth correction conversion parameter thus obtained is a conversion parameter for converting an image for detecting an abnormal point position by the IR camera 30 into an orthographic projection image.

The image conversion unit 120 converts the image of the road surface captured by the digital camera 40 using the third correction conversion parameter, and orthographically converts the image of the digital camera 40 on the road surface at which an abnormal point position is to be detected. Find the projected image. Similarly, an image of the road surface captured by the IR camera 30 is converted by the fourth correction conversion parameter to obtain an orthographic projection image of the image of the IR camera 30 on the road surface at which an abnormal point position is to be detected. The two orthographic projection images obtained in this way are images in which corresponding points can be superimposed on each other.

If the state at the time of photographing the abnormal point position detecting image is substantially the same as the state at the time of calibration,
Without obtaining the third and fourth correction conversion parameters, I
The second correction conversion parameter may be applied to the diagnostic photographed image of the R camera. The correspondence between the image captured by the IR camera 30 and the image captured by the digital camera 40 is determined by the difference between the tilt angle (α1−α2) in addition to the low resolution of the image captured by the IR camera 30. This is difficult because there are differences in image distortion based on the differences in image distortions based on the differences in the focal lengths of the photographing lenses of the cameras, and differences in the size of the images based on the differences in the sizes of the imaging elements of the cameras.

However, since the two images converted into the orthographic projection images are images in which corresponding points can be superimposed on each other as described above, by comparing the two images, the IR camera 30 can be used. An abnormal spot on the road surface found from the image according to the above can be easily specified on the road surface. Further, the image obtained by correcting and converting the image obtained by the IR camera 30 can be inserted into a corresponding position of the image obtained by correcting and converting the image obtained by the digital camera 40 and synthesized. Furthermore, a panoramic image can be created from a plurality of images in which a part of the shooting range overlaps.
Such processing is realized by using the personal computer as an image synthesizing unit. By doing so, it is possible to easily grasp the whole image of the object for which an abnormal point is to be detected and compare it with the drawing, and the above-mentioned specific work becomes easier.

Each of the correction conversions using the correction conversion parameters described above is a kind of coordinate conversion, and its algorithm is known. Next, the procedure of calibration, correction conversion parameter determination, and image conversion processing performed by the processing unit 100 will be described with reference to the flowchart of FIG. 6 illustrating the procedure of calibration and FIG. 7 illustrating the procedure of detecting an abnormal point position. . In FIG.
When the calibration process is started, an image of the digital camera 40 including the calibration reference point is read into the first correction conversion parameter determination unit 110A and displayed (Step 11). Then, the user designates the position of the reference point on the displayed screen and inputs the actual dimensions (step 12). Then, the first correction conversion parameter determination unit 110A calculates a first correction conversion parameter that causes the dimension between the designated reference points to correspond to the input dimension (step 13).

Next, the image of the IR camera 30 including the reference point for calibration is read into the second conversion parameter determination section 110B and displayed (step 14). Then, the position of the reference point on the displayed screen is designated (step 15). Further, the user instructs the type of image of the reference digital camera, that is, whether the image has image distortion due to tilt shooting or has been converted to an orthographic projection image (step 16).

Then, the second correction conversion parameter determination unit 110B sets the second correction conversion parameter so that the position of the reference point specified in step 15 corresponds to the position of the reference point in the image specified in step 16. Is calculated (step 17), the first and second correction conversion parameters are stored in the memory (step 18), and the calibration process is terminated.

In FIG. 7, when the abnormal point position detecting process is started, an image of the digital camera 40 photographing the object whose abnormal point position is to be detected is read into the third correction conversion parameter determination section 110C and displayed. (Step 2
1). Then, the user designates the position of the reference point on the displayed screen and inputs the actual dimensions (step 22). Then, third correction conversion parameter determination section 110C causes third dimension between the designated reference points to correspond to the input dimension.
Is calculated (step 23). Then, a difference between the first correction conversion parameter read from the memory and the third correction conversion parameter obtained in step 23 is obtained (step 24).

Next, the fourth correction conversion parameter determination unit 110D reads the second correction conversion parameter from the memory, corrects the second correction conversion parameter based on the difference between the correction conversion parameters obtained in step 24, and 4 are obtained (step 25). Then, the image conversion unit 120 corrects and converts the digital camera image and / or the IR camera image of the abnormal point position detection target object using the third and / or fourth correction conversion parameters, and performs an orthographic projection image. And display it (Step 2
6) If necessary, an inset synthesis or a panorama synthesis process is performed using an orthographic image based on the image of the digital camera and / or the image of the IR camera obtained in step 26 (step 27), and the position of an abnormal point is determined. The detection processing ends.

The field of application of the present invention is not limited to objects spreading in the depth direction, for example, observation / analysis of road surface conditions, but planar objects facing cameras, for example, piers and building walls and window frames. The present invention can be applied to a planar portion such as a structure of a plant or the like.

[0043]

According to the present invention, the low resolution of the image obtained by the IR camera can be sufficiently compensated for by using the image of a normal camera, and an image can be obtained in which an abnormality such as a road surface can be analyzed in detail. As a result, it is possible to easily understand the abnormal state of the road surface or the like, specify the abnormal place, and analyze the state of the road surface or the like easily.

[Brief description of the drawings]

FIG. 1 is a diagram showing an outline of a road surface photographing state by an image photographing unit according to an embodiment of the present invention.

FIG. 2 is an explanatory diagram illustrating a state of an image captured by the digital camera in FIG. 1;

FIG. 3 is a schematic configuration diagram illustrating a configuration of a processing unit according to the embodiment of the present invention.

FIG. 4 is an explanatory diagram illustrating the principle of a method of correcting tilting distortion.

FIG. 5 is an image of a road surface including a center line as an example of an abnormal point position detection image.

FIG. 6 is a flowchart showing a procedure of a calibration process.

FIG. 7 is a flowchart illustrating a procedure of an abnormal point position detection process.

[Explanation of symbols]

30 IR camera 40 Digital camera 44a to 44d Reference point image in calibration 44a1 to 44d1 Corrected orthographic projection Reference point position 50, 51 in the image Center line image 100 Processing unit 110 Correction conversion parameter determination unit 120 ……………………………………………………………………………………………………………………………………………… 120 …… Image conversion unit 130 ……………… Memory

──────────────────────────────────────────────────の Continued on the front page (51) Int.Cl. ⁶ Identification code FI H04N 7/18 H04N 7/18 EK // E01C 23/01 E01C 23/01

Claims (7)

[Claims] 1. An imaging unit comprising an infrared camera and a normal camera which is installed at a predetermined relative position and shoots a shooting range including a range common to both with a tilt angle, an image of the infrared camera and a normal camera And a processing unit that processes the first image of the infrared camera that has captured a plurality of reference points whose relative positions within the common shooting range are known, and a processing unit of the normal camera. First
Out of the images, the first image of the normal camera having a distortion of the image generated by the tilt shooting based on the position of the reference point in the first image shot by the normal camera. A first conversion parameter determination unit for obtaining a first correction conversion parameter for performing correction conversion to an orthographic projection image; and a first image captured by the infrared camera having an image distortion generated by the tilt shooting A calibration unit comprising: a second conversion parameter determination unit that determines a second correction conversion parameter for correcting and converting to an orthographic projection image based on an image based on shooting by the normal camera. A second image of the infrared camera and the normal camera taken so as to include a plurality of points whose relative positions are known in the common shooting range of the infrared camera and the normal camera. Out of the second image, based on the position of the relative position known point in the second image captured by the normal camera, based on the position of the normal camera having an image distortion caused by the tilt shooting. A third conversion parameter determining unit for obtaining a third correction conversion parameter for correcting and converting the second image into an orthographic projection image; and a difference between the second correction conversion parameter and the third correction conversion parameter. And a fourth conversion parameter determination unit for obtaining a fourth correction conversion parameter for correcting and converting the second image of the infrared camera into an orthographic projection image based on the image correction unit. Abnormal point position detection system by image processing. 2. The image according to claim 1, wherein the calibration unit obtains the second correction conversion parameter by using an image based on an image captured by a normal camera and having an image distortion generated by the tilting image capturing. An abnormal point position detection system using image processing, characterized in that: 3. An orthographic image converted by the first correction conversion parameter according to claim 1, wherein an image based on photographing with a normal camera used by the calibration unit to obtain the second correction conversion parameter is used. An abnormal point position detection system by image processing, which is a projection image. 4. The image correction unit according to claim 1, wherein the image correction unit corrects and converts the second image of the infrared camera with the fourth correction conversion parameter and / or converts the second image of the infrared camera. An abnormal point position detection system by image processing, further comprising: an image conversion unit that corrects and converts the second image using the third correction conversion parameter to obtain an orthographic projection image. 5. The fourth correction conversion according to claim 1, wherein an orthographic projection image of a second image of a normal camera converted by using the third correction conversion parameter is used. An image combining unit that creates an image in which the orthographic projection image of the second image of the infrared camera converted using the parameter is replaced with a corresponding part of the orthographic projection image of the second image of the normal camera. An abnormal point position detection system based on image processing, which is provided. 6. The image processing method according to claim 1, wherein the plurality of reference points whose relative positions are known are structures provided on or adjacent to a road surface. Abnormal point position detection system. 7. A relative position within the common photographing range, which is provided at a predetermined relative position, and which is constituted by an infrared camera and a normal camera that photograph a photographing range including a range common to both with a tilt angle and a normal camera. Capturing a plurality of known reference points as a first image of the infrared camera and a first image of the normal camera; and determining the reference points in the first image captured by the normal camera. Based on the position, a first image captured by the normal camera having an image distortion caused by the tilt image capturing
A first conversion parameter determination step for obtaining a first correction conversion parameter for correcting and converting the image into an orthographic projection image; and A second conversion parameter determining step of obtaining a second correction conversion parameter for correcting and converting the first image into an orthographic projection image based on an image based on shooting by the normal camera; Capturing a second image of the infrared camera and a second image of the normal camera such that the common shooting range of the infrared camera and the normal camera includes a plurality of points whose relative positions are known. Based on the position of the relative position known point in the second image photographed by the normal camera, the communication having the image distortion generated by the tilt photographing. A third conversion parameter determining step for obtaining a third correction conversion parameter for correcting and converting the second image of the camera into an orthographic projection image; and the second correction conversion parameter and the third correction conversion parameter A fourth conversion parameter determining step of obtaining a fourth correction conversion parameter for correcting and converting the second image of the infrared camera into an orthographic projection image based on the difference between the first image and the second image. Abnormal point position detection method.