JP2012159328A - Image processing method, image processing device, and underwater checkup apparatus mounted with the same - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an image processing method, an image processing device and an underwater checkup apparatus mounted with the same that can correct oscillation of and stabilize images by using only image information, and contribute to operating ease of underwater checkup apparatuses.SOLUTION: A subject is consecutively photographed with photographic means mounted on a mobile body; an element of the subject shot in common in two consecutive images is defined to be the feature quantity item; the two consecutively shot images are subjected to image correlation processing to determine the degree of near identity in the feature quantity item; if the nearness to identity fails to reach a prescribed level, a new element of the subject is defined to be the feature quantity item; the position of the feature quantity item in consecutively shot images is fixed to a prescribed position in a one-way direction of the two-dimensional shot images; and the variation quantity in the other direction of the two-dimensional shot images is set to be the shifting extent of the photographic means.

Description

  The present invention relates to an image processing method, an image processing apparatus, and an underwater inspection apparatus having the image processing apparatus for processing an image captured by an imaging apparatus mounted on an underwater moving body. In particular, an image processing method suitable for use in a swimming-type underwater inspection apparatus for visually inspecting in-reactor equipment such as a jet pump in addition to a shroud and a pressure vessel in a nuclear reactor, and an underwater inspection equipped with the image processing apparatus Relates to the device.

  As a prior art related to captured image processing of a camera mounted on a moving body, first, as disclosed in Patent Document 1, a technique for stabilizing the visibility of a camera worn by a human on the head is known.

  Secondly, as disclosed in Patent Document 2, a motion vector is detected from a captured image, and only the captured image is used to suppress image blurring.

  On the other hand, Patent Document 3 is known as a device in which a motion detection sensor is mounted on an underwater inspection apparatus and is steered in an arbitrary direction.

JP 2000-97637 A JP-A-5-219420 JP 2006-224863 A

  The thing of patent document 1 is related with the apparatus which stabilizes the visibility of the camera with which a human body mounts to a head, and detects a motion using the mounted gyro sensor. Moreover, in the thing of patent document 2, a motion is detected by the correlation process of an image. As described above, the devices described in Patent Document 1 and Patent Document 2 detect image shake and correct only the amount of shake in order to stabilize the image.

  However, the underwater inspection apparatus that is the subject of the present invention not only shakes the apparatus, but also may intentionally change the attitude of the apparatus, so it is necessary to follow the image in the opposite direction. The technique described in Patent Document 2 cannot be applied to an underwater inspection apparatus.

  Further, what is described in Patent Document 3 is a device that controls the posture of the underwater inspection device by detecting the posture using a posture detection sensor such as a gyro and performing feedback control, and stabilizes only the image. It is not a thing.

  An object of the present invention is to provide an image processing method, an image processing apparatus, and an underwater inspection apparatus equipped with the image processing method, which can correct image shake using only image information, stabilize the image, and improve the operability of the underwater inspection apparatus. Is to provide.

  In order to achieve the above object, the image processing method of the present invention obtains a photographed image by continuously photographing a subject by a photographing means mounted on a moving body, and commonly photographs two consecutively obtained images. When an element of a subject that has been used is defined as a feature amount, image correlation processing is performed on two images obtained in succession to determine the degree of coincidence of the feature amount, and a degree of coincidence exceeding a predetermined value cannot be obtained Sets the element of a new subject as a feature quantity, fixes the position of the feature quantity in a continuously obtained photographed image to a predetermined position in one direction of the two-dimensional photographed image, and the other direction of the two-dimensional photographed image Is the amount of movement of the photographing means.

  In addition, the position of the feature amount in the continuously obtained captured images is set as a direction orthogonal to the operation direction of the imaging unit as one direction when fixing to a predetermined position in one direction of the two-dimensional captured image.

  In addition, in order to fix the position of the feature amount in continuously obtained captured images to a predetermined position in one direction of the two-dimensional captured image, the feature amount is used to obtain a position between consecutively obtained captured images. An image shake amount is calculated, an image shake amount average is calculated by a moving average of image shake amounts for a preset shooting time, and a corrected image obtained by moving the acquired image at the time by a pixel corresponding to the image shake amount average calculate.

  In addition, in order to determine the feature of a subject that is photographed in common in two images obtained in succession as a feature amount, binarization that converts the image obtained by the photographing means into two gradations of white and black Processing, clustering processing for classifying the aggregate into one of white and black, and area area calculation processing for calculating the area of the aggregate, and a portion having the largest area area is defined as a feature amount.

  In order to achieve the above object, an image processing apparatus of the present invention is an image processing apparatus for acquiring a captured image by continuously shooting a subject by a shooting unit mounted on a moving body and obtaining a moving amount of the shooting unit. The first means for defining the feature of the subject photographed in common in two images obtained in succession as the feature amount, and performing image correlation processing on the two images obtained in succession A second means for determining the degree of coincidence of the feature quantity; a third means for continuously determining a new subject element as the feature quantity when the degree of coincidence at a predetermined level or higher cannot be obtained in the second step; A fourth means for fixing the position of the feature amount in the captured image to a predetermined position in one direction of the two-dimensional captured image, and using the amount of change in the other direction of the two-dimensional captured image as the movement amount of the photographing means; Prepare.

  The one direction in the fourth means is a direction orthogonal to the operation direction of the photographing means.

  Further, in order to fix the position of the feature amount in the captured image continuously obtained by the fourth means to a predetermined position in one direction of the two-dimensional captured image, the fourth means uses the feature amount, Fifth means for calculating an image shake amount between continuously obtained captured images, sixth means for calculating an average image shake amount based on a moving average of the image shake amounts for a preset shooting time, and the time And a seventh means for calculating a corrected image obtained by moving the acquired image by a pixel corresponding to the average image shake amount.

  Further, in the first means, in order to determine the feature of the subject photographed in common in two images obtained in succession as the feature amount, the first means sets the image obtained by the photographing means as white. A binarization processing means for converting to two gray levels of black; a clustering processing means for classifying into an aggregate of either white or black; and an area area calculation processing means for calculating the area of the aggregate. The portion with the largest is defined as a feature amount.

  In order to achieve the above object, an image processing apparatus of the present invention is an image processing apparatus that obtains a processed image in which image shake is suppressed with respect to an acquired image captured by an imaging unit mounted on a moving body, A feature amount calculating means for extracting a feature amount from the image, an image correlation means for calculating an image shake amount of the acquired image at the time and the acquired image at the immediately preceding time using the feature amount registered by the feature amount calculating means, Means for calculating an average image shake amount based on a moving average of image shake amounts for a set time; means for calculating a corrected image by moving an acquired image at the time by a pixel corresponding to the average image shake amount; Means for displaying.

  In order to achieve the above object, the underwater inspection apparatus of the present invention is preset from an inspection vehicle having a driving mechanism capable of swimming in three dimensions and an imaging means capable of visually recognizing a structure in the water, and an image taken by the imaging means. An underwater inspection apparatus having an image processing means for cutting out an image of an image display size and an image display means for displaying an image whose shake has been corrected by the image processing means, wherein the image processing means obtains a feature value from the acquired image. A feature amount calculation unit to extract, an image correlation unit to calculate an image shake amount of the acquired image at the time to be displayed using the feature amount registered by the feature amount calculation unit, and an acquired image at the immediately previous time, and a preset amount of time Means for calculating an average image shake amount based on a moving average of image shake amounts, and means for calculating a corrected image by moving the acquired image at the time by a pixel corresponding to the average image shake amount.

  According to the present invention, it is possible to correct image shake by using only image information, stabilize the image, and improve the operability of the underwater inspection apparatus.

The flowchart which shows the processing content of the image display method by an underwater inspection apparatus. The figure which shows the equipment arrangement | positioning at the time of the underwater inspection work using an underwater inspection apparatus. The bird's-eye view of inspection vehicle 8 used for an underwater inspection device. The figure which showed the definition of the image shake amount in an underwater inspection apparatus. The figure which represented the shake of the image on the time axis. The figure which shows the image of a display image range. The figure explaining the extraction method of an initial feature-value. The figure which shows an underwater camera as an underwater inspection apparatus. The figure which shows the to-be-photographed object used for the conceptual explanation of a subroutine process. The figure which shows the image captured by the camera at each timing. The figure which shows the image which corrected the image blurring amount of the image of FIG.

  Hereinafter, the configuration and operation of an underwater inspection apparatus according to an embodiment of the present invention will be described with reference to the drawings. The underwater inspection apparatus according to the present embodiment will be described with reference to an example used when inspecting defects in a nuclear reactor, particularly when a structure is visually inspected. However, the underwater inspection apparatus is not limited to a nuclear reactor and can be widely applied.

  First, with reference to FIG. 2, an equipment arrangement during an underwater inspection operation using the underwater inspection apparatus according to the present embodiment will be described.

  In the nuclear reactor 1, there are structures such as a shroud 2, an upper lattice plate 3, a core support plate 4, and a shroud support 5. These structures are in the water. In addition, an operation floor 6 that is a work space is provided in the upper part of the nuclear reactor 1, and a fuel changer 7 is also provided in the upper part.

  The structure placed in the water in the nuclear reactor 1 needs to be inspected at an appropriate time, but since it is not an environment in which workers can approach, various work by remote operation by lowering the inspection vehicle 8 into the water. Execute.

  Specifically, the inspection vehicle 8 that has entered the water in the nuclear reactor 1 is connected to the control device 10 on the operation floor 6 via the vehicle cable 9. The control device 10 supplies power for moving the inspection vehicle 8 in the water and navigating it, and performs video communication in order to perform a visual inspection at the inspection target location. Further, a display device 11 is connected to the control device 10 and displays an image from an imaging means mounted on the inspection vehicle 8. Further, a controller 12 is connected to the control device 10 and is operated by a vehicle operator 13a. On the fuel changer 7, the operation assistant 13b runs the vehicle cable 9.

  The control device 10 includes a display for displaying, on the display device 11, a position control means for controlling the movement of the position of the inspection vehicle 8 and an image picked up by the image pickup means mounted on the inspection vehicle 8. Image display processing means for generating an image.

  Next, the configuration of the inspection vehicle 8 used in the underwater inspection apparatus according to the present embodiment will be described with reference to FIG. FIG. 3 is a bird's-eye view showing the configuration of the inspection vehicle 8.

  The inspection vehicle 8 is equipped with a camera unit 20 as an imaging means at the front. The camera unit 20 houses a camera and illumination. The vehicle operator 13a shown in FIG. 2 is configured to be able to operate the movement of the inspection vehicle 8 while confirming the image from the camera unit 20.

  Further, the inspection vehicle 8 is equipped with a lifting thruster 21 for moving up and down, a propulsion thruster 22 for moving back and forth, and translational turning thrusters 23a and 23b for moving and turning left and right. ing. The translation turning thrusters 23a and 23b are rotated in the same direction when moving left and right, and rotated in the opposite direction when turning.

  Next, the concept of the image display method by the underwater inspection apparatus according to the present embodiment will be described with reference to FIGS.

FIG. 4 shows the definition of the image shake amount in the underwater inspection apparatus. FIG. 4 is an image in which an image captured by the camera unit 20 is taken into the control device 10. In the captured image 30, it is assumed that one feature point 31 at time T _i-1 has moved to the position of the feature point 32 at the next time T _i . At this time, the movement amount (a, b) of a in the X direction and b in the Y direction is defined as an image blur amount 33. This image shake amount 33 is calculated by image correlation processing described later, and (a, b) indicates the shake amounts in the X direction and Y direction of the image.

  FIG. 4 illustrates changes in feature points between two images on a time series. Considering a video that is a set of a plurality of continuous images, the feature points are continuous shakes. The cause of the shake of the feature point can be divided into a change in direction intended by the operator and an unintended shake. The direction change intended by the operator corresponds to, for example, a case where the viewpoint of the camera or the underwater inspection apparatus is moved in the X-axis direction in order to monitor the X-axis direction. As the unintended vibration, for example, a vibration due to a collision with a structure inside the nuclear reactor, a water wave caused by the vibration of the structure being propagated in water, or the like can be assumed.

  FIG. 5 is a diagram showing image shake on the time axis. In FIG. 5, the horizontal axis indicates time, and the vertical axis indicates the total shake amount from the initial image. Here, the total vertical shake amount is, for example, the shake amount in the X direction in FIG. It is shown. Accordingly, there is actually another coordinate axis that shows the amount of shake in the Y direction with time, but here, the amount of shake in the X direction will be described as a representative example.

  40 is a temporal trace of the amount of shake in the X direction, and this is the movement trajectory of the observed feature point in the X direction. The movement trajectory is considered to have some frequency component because it increases and decreases as appropriate. On the other hand, when the operator moves the viewpoint of the camera or the underwater inspection apparatus in the X-axis direction in order to monitor the X-axis direction, the amount of movement in the X-axis direction per unit time is considered to be substantially the same. . That is, it should appear as a linear change on the illustration of FIG. For this reason, when viewed as a movement on the time axis, the direction change intended by the operator is considered to have a low frequency, and an unintended shake is considered to have a high frequency.

In the present invention, in order to grasp only the direction change intended by the operator, unintended shake is excluded. Therefore, paying attention to the movement trajectory in FIG. 5, the change 40 in the image shake amount has a large vibration as shown by the solid line. Here, a moving time 41 of image shake amounts of 3 samples, for example, a preset time is calculated. As a result of this processing, the high frequency component is removed, and only the change intended by the operator remains. Compared moving average from the time T ₁ of the case not implementing the change amount 42 to T _2, the variation 43 in the case of carrying out the moving average is reduced. This amount is used as the image correction amount 43 at time _{T 2.} The image correction amount 43 is calculated independently of a and b in FIG. 3, that is, the X direction and the Y direction, and the values are A and B, respectively.

FIG. 6 is a diagram showing an image of the display image range. The display image 51 at the next time T _i is displayed so as to be shifted by the image shift amount 52 in the X direction and the image shift amount 53 in the Y direction with respect to the display image 50 at the time T _i−1 . It should be noted that since there is no change in the display size at either time T _i-1 or time T _i , there is no image information corresponding to the shift, and therefore the blank area 54 is inserted at time T _i . .

  Next, processing contents of the image display method by the underwater inspection apparatus according to the present embodiment will be described with reference to FIG. These processes are executed by the image display processing means of the control device 10 shown in FIG. 2, and are flowcharts that embody the methods of FIGS.

  In this flow, after the processing is started in step S00, a time constant N is input in step S01. This time constant indicates the width of the moving average shown in FIG. 5, and is specifically the number of moving average samples.

Next, in step S02, it performs initial image processing at time _{T 0.} In step S03, an initial photographed image is read. In step S04, distortion caused by the camera lens is corrected. In step S05, the initial image is stored. Next, in step S06, in the read elaborate image extracting initial features for the next step at time T _0, and registers.

Here, the extraction method of the initial feature amount will be described with reference to FIG. First, in the initial image 55 of FIG. 7, for example, a, b, and c are photographed as elements of the subject. Since the elements a, b, and c are not uniform in density, binarization is performed on the image 55 acquired at time T ₀ to obtain a uniform gradation image 56, and an image 57 subjected to clustering processing is used to obtain each element An element number is assigned to each area. Then, the area (number of pixels) constituting each element is calculated, and the one having the largest area is set as the feature amount 58. The range to be registered as the feature amount is a rectangular area H that includes the element.

  Next, in step S07, an initial image correlation within a range determined by the time constant N is calculated. Specifically, the initial image correlation calculation process is to call the subroutine of image shake amount calculation in step S09 in step S08 and execute this repetitive calculation. The subroutine of Step S09 is composed of a series of processes from Step S10 to Step S22.

  Since the present invention has a feature in the processing in this subroutine, in order to help understanding of this concept, an example will be described in which the subject in FIG. 9 is photographed. FIG. 9 shows a wall surface to be photographed, which is a camera subject. However, the viewing angle range (shooting range) of the camera is G, and only a part of the wall surface to be shot can be captured by the camera.

It is assumed that characters A, B, and C are drawn at the illustrated positions on the wall surface to be photographed, and the operator moves the camera in the lateral direction (X direction) and performs photographing. It is assumed that the continuous time at this time is T _i−1, T, T _{i + 1} , and the image G taken at each time is G _i−1, G _i , G _{i + 1} . At this time, the operator's intention is to move the camera in the horizontal direction (X direction), but the shooting range including the vertical change (Y direction) is caused by the shaking of the image. Yes.

The subroutine of step S09, first, to acquire the image of the current time _{T i} in step S10. In particular, it captures the image of the current time in step S11, to correct the distortion in step S12, stores the original image _{G i} at step S13. This image is a center of G _i in FIG. 10, the letter A generally centrally, has become a part of B is reflected in the lower right.

Next, in step S14, an image at the previous time T _i-1 is extracted. This image is G _i-1 at the center of FIG. 10, and the letter A is moved almost to the bottom of the center, but B is not yet visible. Further reads the feature quantity 58 registered in step S06 in step S15, the image correlation calculation between the image _{G i-1} of the original image _{G i} and the previous time _{T i-1} of the current time _{T i} in step S16. However, it is assumed that the feature amount 58 recognized between the images is the letter A.

  Here, the image correlation calculation is to determine that a predetermined feature amount (58 in FIG. 7, character A) exists in two time-series images. If the two images have the same predetermined feature amount, the movement of the feature amount between the images can be tracked. If they do not have the same feature amount, it is necessary to set a new feature amount and use it as a new reference for connecting two images.

For this purpose, the feature amount switching determination is performed in step S17 using the feature amount coincidence ratio between the images G _i and G _i-1 at the current time T _i and the previous time T _i-1 calculated by the image correlation calculation. For the determination, a threshold value in the range from 0 to 1 set in advance is used. If the threshold value is equal to or less than the threshold value, it is considered that the predetermined feature value is lost. Therefore, the feature value is updated in step S19. This is stored as the next new predetermined feature amount. Here, the extraction of a new feature amount is the same as the method of FIG. 7 described above. Note that when the feature amount matching rate is equal to or greater than the threshold value, a predetermined feature amount is continuously observed. Therefore, movement tracking using this feature amount is possible, and thus the feature amount is not updated in step S18. .

Here, in obtaining the feature amount matching rate, in the image processing of FIG. 7, the feature amount is set to 58 on the screen of 57 and a rectangular region H including the element of the feature amount 58 is defined. In calculating the feature amount matching rate between the images G _i and G _{i-1 in} FIG. 10, since the rectangular regions H _i and H _i-1 are included in almost all regions, the matching rate is almost 100%. Can be understood. In this regard, in calculating the feature amount matching rate between the images G _i and G _{i + 1} at the next timing, since the rectangular regions H _i and H _{i + 1} are partially missing, the matching rate is 100%. Although it is not necessary, the movement tracking by this feature amount can be continued by comparing with a threshold value in a range from 0 to 1 set in advance.

  9 and 10 show only images at three consecutive timings. However, if the camera is moved in the right direction as it is, it is expected that the feature amount A will deviate from the viewing angle. At this time, the feature amount matching rate for the character A becomes equal to or less than the threshold value. In this case, since a new letter B enters the viewing angle, continuous monitoring is possible by executing the same processing as described above using the letter B as a new feature amount.

  Finally, in step S21, an image shake amount is determined from the movement amounts (a, b) in the X-axis direction and the Y-axis direction shown in FIG. 3, and stored in step S22.

  In step S23, the above process is repeated by incrementing the time. After the completion, the image correction process is performed in step S24.

  Specifically, first, in step 25, in order to calculate the image shake amount at the current time, subroutine step S09 is called, and the processing from step S09 to step S22 is performed.

  Next, in step S26, the past image shake amount stored in step S22 is read. What is read here is the result of the image shake amount at N consecutive times from i-N to i-1 with respect to the current time i. Using the result, the image shake amount average (A, B) of FIG. 6 is calculated in step S27, and the number of pixels of the image shake amount average (A, B) is shifted in step S28. This step is based on the method shown in FIG.

  FIG. 11 shows an image obtained by correcting the image shake amount of the image shown in FIG. In FIG. 10, since the image shake amount is not corrected, the position in the height direction of the character A as the feature amount is not determined, but in FIG. 11 in which the image shake amount is corrected, the height of the character A as the feature amount is determined. Directional position is almost the same. Therefore, if a position change in the horizontal direction (X direction) of the character A as a feature amount is calculated from this image, this can be used as a camera movement amount.

  As is clear by comparing FIG. 11 and FIG. 10, here, the position of the feature amount in the continuously obtained captured images is fixed at a predetermined position in one direction of the two-dimensional captured image, and the two-dimensional The amount of change in the other direction of the photographed image is the amount of movement of the photographing means. In this embodiment, the concept of moving average of shake amount is used for this realization. In addition, the position of the feature amount in the continuously obtained captured images is set as a direction orthogonal to the operation direction of the imaging means as one direction when fixing to a predetermined position in one direction of the two-dimensional captured image.

  The result is displayed as a corrected image in step S29. At this time, as shown in FIG. 6, a blank area is inserted and displayed at the edge of the screen. Finally, loop determination is performed in the next step S30. If there is no end input from the operator in step S31, the process jumps from step S32 to step S33, and the processes from step S23 to step S29 are repeated. If there is an end input from the operator, end processing is performed in step S34.

  As described above, according to the present embodiment, it is possible to correct the shake of the image using only the image information and to follow the intended change in posture. As a result, it is possible to stabilize the captured image mounted on the moving body, and improve operability during movement.

  Moreover, it becomes possible to stabilize the image of the camera mounted in the underwater inspection apparatus, and the inspection efficiency in visual inspection can be improved.

  Next, the configuration and operation of an underwater inspection apparatus according to another embodiment of the present invention will be described with reference to FIG. The arrangement of equipment during underwater inspection using the underwater inspection apparatus according to the present embodiment is shown in FIG. 1 using a submersible camera 59 instead of the inspection vehicle 8 and holding a cable including a cable instead of the vehicle cable. 60 is different, and other configurations, image processing methods, and display methods are the same. The underwater camera 59 is equipped with a camera unit 61 and lamp units 62a and 62b. The video imaged by the camera unit 61 is subjected to image processing in the control device 10 and displayed on the display device 11.

  The present embodiment is intended to achieve the same purpose as the first embodiment, but the holding device has a low up-and-down method unlike the inspection vehicle 8 in which the camera is mounted. Therefore, shakes that are not desired by the operator are mainly horizontal shakes. In this case, in the description of the first embodiment, the same effect can be obtained by correcting only the horizontal direction of the image, that is, the X direction when calculating the image shake amount described in detail with reference to FIG. Also, it is possible to shorten the calculation time required for image processing and improve the responsiveness.

DESCRIPTION OF SYMBOLS 1 ... Reactor 2 ... Shroud 3 ... Upper lattice board 4 ... Core support plate 5 ... Shroud support 6 ... Operation floor 7 ... Fuel changer 8 ... Inspection vehicle 9 ... Vehicle cable 10 ... Control device 11 ... Display device 12 ... Controller 20 ... Camera unit 21 ... Elevating thruster 22 ... Propulsion thrusters 23a and 23b ... Translational turning thruster 30 ... Captured image 31 ... Feature point 32 at time Ti-1 ... Feature point 33 at time Ti ... Image shake amount 40 ... Change in image shake amount 41 ... Moving average 42 of image shake amount ... Image shake amount 43 between times T1 and T2 ... Image correction amount 50 at time T2 ... Display image 51 at time Ti-1 ... Display image 52 at time Ti ... X-direction image shift amount 53 ... Y-direction image shift amount 54 ... Blank area 55 at time Ti ... Time ₀ images acquired 56 ... binarized image 57 ... clustering images 58 ... feature amount 59 ... underwater camera body 60 ... holding jig 61 ... camera unit 62a, 62b ... lamp unit

Claims (10)

  The subject is continuously photographed by the photographing means mounted on the moving body to obtain the photographed image, the element of the subject photographed in common in the two images obtained in succession is set as the feature amount, and continuously Image correlation processing is performed on the two obtained images to determine the degree of coincidence of the feature amounts. If the degree of coincidence is not equal to or higher than a predetermined value, a new subject element is set as the feature amount, and The position of the feature amount in the captured image obtained in this way is fixed at a predetermined position in one direction of the two-dimensional captured image, and the amount of change in the other direction of the two-dimensional captured image is set as the movement amount of the photographing means. A featured image processing method. The image processing method according to claim 1,
The position of the feature amount in the continuously obtained captured images is set to a direction orthogonal to the operation direction of the imaging means as the one direction when fixing to a predetermined position in one direction of the two-dimensional captured image. A featured image processing method. The image processing method according to claim 1 or 2,
In order to fix the position of the feature amount in the continuously obtained captured image at a predetermined position in one direction of the two-dimensional captured image, the image between continuously captured images using the feature amount is used. A corrected image in which a shake amount is calculated, an image shake amount average is calculated by a moving average of the image shake amounts for a preset shooting time, and an acquired image at the time is moved by a pixel corresponding to the image shake amount average. An image processing method characterized by calculating. The image processing method according to any one of claims 1 to 3,
A binarization process for converting an image obtained by the photographing means into two gradations of white and black in order to determine an element of a subject photographed in common in two images obtained in succession as a feature amount; , An image having a clustering process for classifying into an aggregate of either white or black and an area area calculation process for calculating an area of the aggregate, wherein a portion having the largest area area is used as a feature amount Processing method. In an image processing apparatus for obtaining a captured image by continuously photographing a subject with a photographing means mounted on a moving body and obtaining a moving amount of the photographing means,
A first means for determining the feature of an object photographed in common in two images obtained in succession, and the feature amount by performing image correlation processing on the two images obtained in succession A second means for determining the degree of coincidence, and a third means for determining a new subject element as a feature amount when a second degree of coincidence is not obtained in the second step. And a fourth means for fixing the position of the feature amount in the image at a predetermined position in one direction of the two-dimensional photographed image and using the change amount in the other direction of the two-dimensional photographed image as the movement amount of the photographing means. An image processing apparatus. The image processing apparatus according to claim 5.
An image processing apparatus according to claim 4, wherein the one direction in the fourth means is a direction orthogonal to an operation direction of the photographing means. The image processing apparatus according to claim 5 or 6,
In order to fix the position of the feature amount in the captured image continuously obtained by the fourth means to a predetermined position in one direction of the two-dimensional captured image, the fourth means uses the feature amount. Fifth means for calculating an image shake amount between continuously obtained captured images, sixth means for calculating an average image shake amount based on a moving average of the image shake amounts for a preset shooting time, An image processing apparatus comprising: a seventh means for calculating a corrected image obtained by moving an acquired image at a time by a pixel corresponding to the average image shake amount. The image processing apparatus according to any one of claims 5 to 7,
In the first means, in order to determine the feature of the subject photographed in common in two images obtained in succession as the feature amount, the first means determines that the image obtained by the photographing means is white. A binarization processing means for converting to black two gradations, a clustering processing means for classifying into an aggregate of either white or black, and an area area calculation processing means for calculating the area of the aggregate. An image processing apparatus characterized in that a portion having the largest area is used as a feature amount. An image processing apparatus that obtains a processed image in which image shake is suppressed with respect to an acquired image captured by an imaging unit mounted on a moving body,
Feature amount calculating means for extracting feature amounts from the acquired image, and image correlation for calculating an image shake amount between the acquired image at the time displayed using the feature amount registered by the feature amount calculating means and the acquired image at the immediately preceding time Means, a means for calculating an average image shake amount based on a moving average of the image shake amounts for a preset time, and a corrected image is calculated by moving the acquired image at the time by a pixel corresponding to the average image shake amount. An image processing apparatus comprising: means for displaying and means for displaying the corrected image. An inspection vehicle having a driving mechanism capable of swimming in three dimensions and an imaging means capable of visually recognizing a structure in the water, an image processing means for cutting out an image having a preset image display size from the image taken by the imaging means, and An underwater inspection apparatus having image display means for displaying an image whose shake has been corrected by an image processing means,
The image processing unit includes a feature amount calculating unit that extracts a feature amount from the acquired image, an image shake of an acquired image at a time displayed using the feature amount registered by the feature amount calculating unit, and an acquired image at a previous time. Image correlation means for calculating the amount, means for calculating the average image shake amount based on the moving average of the image shake amounts for a preset time, and moving the acquired image at the time by the pixel corresponding to the average image shake amount An underwater inspection apparatus comprising: means for calculating a corrected image, and means for displaying the corrected image.

Images