JP2013134114A - Image processor - Google Patents

Abstract

PROBLEM TO BE SOLVED: To estimate a time when a photographic image is photographed on the basis of the photographic image of a mobile body and the photographic image of a sea wave being a background.SOLUTION: An image processor includes: a separation part 2 for separating verification target image data obtained by photographing a mobile body on a sea surface into a mobile body image indicating the mobile body and a background image including a sea wave other than the mobile body; a scale estimation part 3 for estimating the three-dimensional coordinate of the sea wave in the background image; an FFT processing part 4 for acquiring a sea wave spectrum; a Hasselmans conversion part 5 for performing the Hasselmans conversion of the sea wave spectrum acquired by the FFT processing part 4 and acquiring an estimation synthetic aperture radar image; a measured image storage part 7 for storing multiple measured images obtained by photographing a prescribed area at a predetermined time interval together with each photographic time; and an evaluation comparison part 6 for calculating a similarity degree indicating similarity between the estimation synthetic aperture radar image acquired by the Hasselmans conversion part 5 and each measured image stored in the measured image storage part 7, and estimating the photographic time of verification target image data by using the photographic time of the measured image indicating high similarity.

Description

  The present invention relates to an image processing apparatus that estimates the time at which an image is captured from a captured image of a moving object on the sea surface.

  Marine images from conventional satellite and aircraft-borne synthetic aperture radars are used to provide real-time / quasi-real-time wave information or improve the accuracy of wave prediction numerical models. In addition, the marine image of the synthetic aperture radar is used as an environmental accident situation grasp and legal material (for example, see Non-Patent Document 1).

  On the other hand, a technique is also known in which the ocean wave shape is represented by a spectrum, and this is nonlinearly transformed to create a corresponding synthetic aperture radar image (virtual) (see, for example, Non-Patent Document 2). According to Non-Patent Document 2, a corresponding synthetic aperture radar image is obtained by performing Hasselman transform on the sea surface wave spectrum, and a corresponding sea surface wave spectrum is obtained by performing inverse Hasselman transform on the synthetic aperture radar image. can get.

In addition, a technique for randomly modeling the ocean wave shape for generating a synthetic aperture radar image is also known (see, for example, Non-Patent Document 3).
Furthermore, there are many known methods for recognizing and identifying a specific object such as a face or a ship from the background from a captured image (see, for example, Non-Patent Document 4).

Kazuo Ouchi, Basics of Synthetic Aperture Radar for Remote Sensing, 1.2 Fields of Application of Synthetic Aperture Radar, 1.2.1 Ocean, Tokyo Denki University Press, January 20, 2004 K. Hasselmann et al. , On the non-linear mapping of an ocean wave spectrum into a synthetic approach, radical image spectrum and it's inversion, Journal of Geology. 96, no. C6, pp. 10, 713-10, 729, June 15, 1991) F. Belizzi et al. Sea-wave fractal spectrum for SAR remote sensing, IEEE Proc. -Radar, Sonar Navig. , Vol. I48, no. 2, pp. 56-66, April 2001 Takayoshi Yamashita, Hironobu Fujiyoshi, Features Effective for Specific Object Recognition, Information Processing Society of Japan Research Report CVIM 165 Nov. 2008, pp. 221-236

  However, in the conventional remote sensing technology using synthetic aperture radar disclosed in Non-Patent Document 1 to Non-Patent Document 3 described above, the mutual positional relationship between a moving target such as a ship and the background sea level is different from that of the actual image. The time at which the image was captured by comparing the synthetic aperture radar image at the position where the capturing time is known and the two-dimensional photograph of the ship on the sea surface where the capturing time is unknown by comparing each other with the moving target as a clue. There was a problem that it was difficult to estimate.

  Further, in the conventional algorithm for converting a sea surface wave spectrum and a synthetic aperture radar image, it is only necessary to convert / reverse the sea surface wave spectrum expressed by a mathematical formula and the synthetic aperture radar image to each other. There was a problem that it was difficult to estimate the time when the image was taken.

  The present invention has been made to solve the above-described problems, and an object thereof is to estimate the time at which the photographed image was photographed from the photographed image of the moving body and the photographed image of the sea wave in the background. .

  An image processing apparatus according to the present invention includes: a separation unit that separates verification target image data obtained by capturing a moving object on the sea surface into a moving object image indicating the moving object and a background image including sea surface waves other than the moving object; A scale estimation unit that estimates the three-dimensional coordinates of the sea surface wave in the background image based on the scale of the moving object in the moving body image separated from the unit, and a fast Fourier transform of the three-dimensional coordinates of the sea surface wave estimated by the scale estimation unit An FFT processing unit that acquires the sea surface wave spectrum, a Hasselman conversion unit that acquires the estimated synthetic aperture radar image by performing Hasselman conversion on the sea surface wave spectrum acquired by the FFT processing unit, and a predetermined area in advance. A plurality of measured images captured at intervals are recorded in the measured image storage unit that stores the captured time together with the shooting time, the estimated synthetic aperture radar image acquired by the Hasselman conversion unit, and the measured image storage unit. By the similarity calculating that shows similarity with the measured image is one and a evaluation comparing unit for estimating the shooting time of the verification target image data using the imaging time of the actual image showing high similarity.

  According to the present invention, it is possible to estimate the time when the captured image was captured from the captured image of the moving body and the captured image of the sea wave in the background.

1 is a block diagram illustrating a configuration of an image processing device according to Embodiment 1. FIG. 3 is a flowchart illustrating an operation of the image processing apparatus according to the first embodiment. 6 is a diagram illustrating a similarity table of the image processing apparatus according to Embodiment 1. FIG. 6 is a flowchart illustrating an operation of a similarity calculation unit of the image processing apparatus according to the first embodiment. 6 is a flowchart illustrating an operation of an estimated time calculation unit of the image processing apparatus according to the first embodiment. 6 is a diagram illustrating weighting processing of an estimated time calculation unit of the image processing apparatus according to Embodiment 1. FIG. 6 is a block diagram illustrating a configuration of an image processing device according to Embodiment 2. FIG. 6 is a flowchart illustrating an operation of the image processing apparatus according to the second embodiment. 10 is a flowchart illustrating an operation of a Hasselman inverse transform unit of the image processing apparatus according to the second embodiment. 10 is a flowchart illustrating an operation of a similarity calculation unit of the image processing apparatus according to the second embodiment.

Embodiment 1 FIG.
FIG. 1 is a block diagram showing a configuration of an image processing apparatus according to Embodiment 1 of the present invention. In FIG. 1, the image processing apparatus 10 includes a reading unit 1, a separation unit 2, a scale estimation unit 3, an FFT processing unit 4, a Hasselman conversion unit 5, an evaluation comparison unit 6, and a measured image storage unit 7. In the first embodiment, a case where a moving body such as a ship on the sea surface is photographed will be described as an example.
The reading unit 1 converts a captured image to be verified (hereinafter referred to as a verification target image) into electronic data. As a method for converting to electronic data, for example, there is a method in which a photographed image printed on a paper medium is divided into two-dimensional pixels, and the density of each pixel is held in a table. Besides this method, a general electronic data conversion method can be applied. Note that when a captured image is input as electronic data instead of a captured image printed on a paper medium, only the input of electronic data is accepted.

  The separation unit 2 separates the electronic data of the verification target image into a moving body image indicating a moving body such as a ship and a background image including a sea wave other than the moving body. For the separation of the moving body image and the background image, a method such as pixel difference using the fact that the luminance of adjacent pixels is different between the sea wave of the moving body and the background is applied. Details of the image recognition method used for separation of the moving body and the sea wave of the background are described in Non-Patent Document 4 and the like described above. The scale estimation unit 3 estimates the type (for example, ship type) and size of the moving object from the shape of the moving object in the moving object image input from the separation unit 2, and sets the estimated type and size of the moving object. Based on this, the size of the sea wave in the background image is estimated to generate a three-dimensional image of the sea wave.

  The FFT processing unit 4 performs a fast Fourier transform (FFT) process on the three-dimensional image of the sea surface wave to calculate a sea surface wave spectrum. The Hasselman transform unit 5 performs Hasselman transform on the sea surface wave spectrum input from the FFT processing unit 4 to obtain an estimated synthetic aperture radar image. Note that details of the Hasselman transform are disclosed in Non-Patent Document 3 and the like, and thus description thereof is omitted. The evaluation comparison unit 6 compares the estimated synthetic aperture radar image input from the Hasselman conversion unit 5 with the actual measurement image stored in the actual measurement image storage unit 7, and the verification target image read by the reading unit 1 is captured. The estimated date and time are estimated and output as estimated date and time information.

The measured image storage unit 7 captures electronic data (hereinafter referred to as measured image data) of a photographed image obtained by photographing a predetermined space at a predetermined time interval τ (for example, every two hours) in advance. And location information. Stored together with time information indicating the time. Although FIG. 1 shows a configuration in which the measured image storage unit 7 is provided in the image processing apparatus 10, it may be configured as a storage unit outside the image processing apparatus 10.
In the first embodiment, as shown in FIG. 1, the user is provided with information about the date when the verification target image was taken and a rough shooting location, and the candidate area L and the candidate time zone T are determined from the provided information. It is set to estimate the shooting time of the verification target image.

FIG. 2 is a flowchart showing the operation of the image processing apparatus according to the first embodiment.
The reading unit 1 reads the verification target image and converts it into electronic data (step ST1).
The separation unit 2 separates the electronic data of the verification target image converted in step ST1 into a moving body image and a background image (step ST2). The scale estimation unit 3 estimates the type and size of the moving object from the moving object shape of the moving object image separated in step ST2, and estimates the scale of the verification target image (step ST3). Furthermore, the scale estimation unit 3 generates a three-dimensional image of sea surface waves based on the scale of the verification target image estimated in step ST3 and the background image separated in step ST2 (step ST4). The FFT processing unit 4 performs a fast Fourier transform on the three-dimensional image of the sea surface wave, and acquires a sea surface wave spectrum (step ST5). The Hasselman transform unit 5 performs Hasselman transform on the sea surface wave spectrum and acquires an estimated synthetic aperture radar image (step ST6). The evaluation comparison unit 6 compares the estimated synthetic aperture radar image acquired in step ST6 with the actual measurement image satisfying the candidate region L and the candidate time zone T stored in the actual measurement image storage unit 7, and the verification target The time when the captured image was captured is estimated (step ST7), the estimated time is output as estimated date and time information (step ST8), and the process is terminated.

Next, details of the evaluation comparison unit 6 will be described.
As shown in FIG. 1, the evaluation comparison unit 6 includes a similarity calculation unit 61, a similarity table storage unit 62, and an estimated time calculation unit 63.
Information on the candidate region L and the candidate time zone T is input to the similarity calculation unit 61 from the outside. The similarity calculation unit 61 reads a plurality of actually measured images that satisfy the conditions of the input candidate region L and candidate time zone T from the actually measured image storage unit 7, and inputs the read actual images and the Hasselman conversion unit 5. A similarity indicating similarity to the estimated synthetic aperture radar image is calculated. A typical calculation method of similarity is a method of obtaining a correlation for each two-dimensional pixel, but other general methods may be applied. Further, the similarity calculation unit 61 creates a similarity table in which the actual image capturing time and the calculated similarity are associated with each other, and stores the similarity table in the similarity table storage unit 62.

FIG. 3 is a diagram illustrating an example of a similarity table stored in the similarity table storage unit of the image processing apparatus according to the first embodiment.
The similarity table is a table showing the similarity between the estimated synthetic aperture radar image and the measured image at time t at a position c1 on a certain date. In the example of FIG. 3, the similarities corresponding to 6:00, 8:00,..., 20:00 are stored at time t. The time t is the time when the actual measurement image that satisfies the condition of the candidate time zone T is captured, and indicates that the image is captured at a predetermined time interval τ. In the example of FIG. 3, the degree of similarity with the actually measured image taken at the time interval τ = 2: 00 is calculated. The similarity indicates a correlation value (0 ≦ similarity ≦ 1) between the estimated synthetic aperture radar image and the actually measured image. When similarity = 0, there is no correlation, and when similarity = 1, the correlation is strong. It is shown that.
In the example of FIG. 3, a similarity table in which the time t and the similarity are associated with each other is shown, but other information such as date information and position information can also be associated and stored.

The estimated time calculation unit 63 refers to the similarity table stored in the similarity table storage unit 62, selects the time t1 at which the similarity is the maximum and the time t2 at which the similarity is the second largest, and the selected time The photographing time of the verification target image is estimated based on t1 and t2.
When referring to the similarity table shown in FIG. 3A, time 12:00 is selected as time t1, and time 14:00 is selected as time t2. Similarly, when referring to the similarity table shown in FIG. 3B, time 18:00 is selected as time t1, and time 6:00 is selected as time t2.

Next, the operation of the evaluation comparison unit 6 will be described with reference to FIGS.
FIG. 4 is a flowchart showing the operation of the similarity calculation unit of the image processing apparatus according to the first embodiment.
When the estimated synthetic aperture radar image is input from the Hasselman conversion unit 5 (step ST11), the similarity calculation unit 61 acquires the actual measurement image at time t and position c from the actual measurement image storage unit 7 (step ST12). Next, the similarity between the estimated synthetic aperture radar image acquired in step ST11 and the actually measured image acquired in step ST12 is calculated (step ST13), and the calculated similarity is associated with the time t indicating the shooting time of the actually measured image. Is written in the similarity table (step ST14). Thereafter, it is determined whether or not the similarity has been calculated for all the times t within the candidate time during which the captured image to be verified is captured (step ST15). When the similarity is not calculated for all times t (step ST15; NO), the time interval τ is added to the time t (step ST16), and the process returns to step ST12. On the other hand, when the similarity is calculated for all times t (step ST15; YES), the similarity table is stored in the similarity table storage unit 62 (step ST17), and the process is terminated.

FIG. 5 is a flowchart showing the operation of the estimated time calculation unit of the image processing apparatus according to the first embodiment.
The estimated time calculation unit 63 refers to the similarity table stored in the similarity table storage unit 62, and reads all similarities in the candidate time zone T (step ST21). Of the similarities read in step ST21, a time t1 at which the similarity is the maximum is selected (step ST22), and a time t2 having the second highest similarity is selected (step ST23). Subsequently, the estimated time calculation unit 63 determines whether or not the time t1 selected in step ST22 and the time t2 selected in step ST23 are adjacent times, that is, the difference between the time t1 and the time t2 is the time interval τ. It is determined whether or not (step ST24).

  When the time t1 and the time t2 are adjacent times (step ST24; YES), it is assumed that there is a time when the similarity is “1” between the time t1 and the time t2, and the similarity between the time t1 and the time t2 The time tw that is weighted between the time t1 and the time t2 is calculated based on the degree difference (step ST25), the calculated time tw is output as the estimated date and time information (step ST26), and the process ends. On the other hand, when the time t1 and the time t2 are not adjacent times (step ST24; NO), the binary values of the time t1 and the time t2 are output as estimated date and time information (step ST27), and the process is terminated.

The operation shown in the flowchart of FIG. 5 will be described using a specific example of the similarity table shown in FIG.
First, the case of referring to the similarity table of FIG.
The estimated time calculation unit 63 selects the time 12:00 having the maximum similarity “0.9” as the time ST1 as steps ST22 and ST23, and the time 14 having the second largest similarity “0.7”: 00 is selected at time t2. Step ST24; YES, since it is determined that the time interval between time t1 = 12: 00 and time t2 = 14: 00 is τ = 2: 00, the process proceeds to step ST25. As step ST25, the time tw = 12: 30 is calculated by weighting between the time t1 and the time t2 with the difference between the similarity (t1) = 0.9 and the similarity (t2) = 0.7. The calculation of the time tw is shown in FIG. The time tw calculated as step ST26 is output as estimated date and time information.

Next, the case of referring to the similarity table of FIG.
The estimated time calculation unit 63 selects time 18:00 having the maximum similarity “0.9” as the time ST1 as steps ST22 and ST23, and time 6 having the second largest similarity “0.8”: 00 is selected at time t2. Step ST24; NO, since it is determined that the time interval between time t1 = 18: 00 and time t2 = 6: 00 is not τ = 2: 00, time t1 = 18: 00 and time t2 = 6 as step ST26. : 00 binary value is output as estimated date and time information.

  As described above, according to the first embodiment, the scale of the verification target image is estimated from the shape of the moving body of the moving body image, and a three-dimensional sea surface wave is generated from the background image based on the estimated scale. The scale estimation unit 3, the FFT processing unit 4 for acquiring the sea surface wave spectrum, the Hasselman conversion unit 5 for acquiring the estimated synthetic aperture radar image from the sea surface wave spectrum, the estimated synthetic aperture radar image and the measured image are compared, Since the evaluation comparison unit 6 that estimates the time at which the verification target image is captured is provided, the time at which the verification target image is captured can be estimated from the sea wave of the moving body image and the background image. Specifically, a synthetic aperture radar image with a known photographing time and a two-dimensional image of a moving body on the sea surface with an unknown photographing time are compared with each other based on the moving object, and the photographing time of the two-dimensional image is obtained. Can be estimated.

  Further, according to the first embodiment, the similarity indicating the similarity between the plurality of actually measured images and the estimated synthetic aperture radar image is calculated, and the calculated similarity and the photographing time of the actually measured image are shown in association with each other. The similarity calculation unit 61 that creates the similarity table, the time t1 at which the degree of similarity is the maximum, and the time t2 at which the degree of similarity is the second largest are selected, and the verification target image is captured based on the selected times t1 and t2. Since it is configured to include the estimated time calculation unit 63 that estimates the time, it is possible to estimate the time when the verification target image was captured based on the similarity of the images.

  As described above, according to the first embodiment, when estimating the time when the sea-like moving body is captured using the sea wave of the background image, the synthetic aperture radar image uses the moving object such as a ship and the background. The correct shooting time can be calculated even when the mutual position of the image appears to deviate from the actual position. As a result, it is judged whether or not the accident situation photo and the report information (time, place, etc.) submitted by the ship are correct, as well as a judgment document for ship accidents, marine accident trials, compensation for environmental damage, etc. or reference materials. be able to.

  In the first embodiment described above, the configuration in which the information regarding the candidate region L and the candidate time zone T is input to the evaluation comparison unit 6 from the outside has been described. However, such information is not input, and the measured image storage unit 7 may be configured to compare all the measured images stored in 7 and the estimated synthetic aperture radar image to calculate the similarity. In addition, when the verification target image is input as electronic data, the candidate time zone T may be set from the shooting time included in the electronic data. Thus, the setting method of the candidate area | region L and the candidate time slot | zone T can be changed suitably.

Embodiment 2. FIG.
In the first embodiment described above, the configuration for performing the Hasselman transform on the sea surface wave spectrum acquired from the verification target image is shown. However, in the second embodiment, the actual measurement image stored in the actual measurement image storage unit 7 is displayed. The structure which performs Hasselman reverse transformation is shown.
FIG. 7 is a block diagram illustrating a configuration of the image processing apparatus according to the second embodiment. In the following description, the same or corresponding parts as the components of the image processing apparatus 10 according to the first embodiment are denoted by the same reference numerals as those used in the first embodiment, and description thereof is omitted or simplified.

First, the reading unit 1 to the FFT processing unit 4 have the same configuration as that of the first embodiment. In the second embodiment, the sea surface wave spectrum generated by the FFT processing unit 4 is referred to as a verification target sea surface wave spectrum. This sea surface wave spectrum to be verified is output to the similarity calculation unit 61 ′ of the evaluation comparison unit 6 ′.
When the reading information indicating the reading of the verification target image is input from the reading unit 1, the Hasselman inverse transformation unit 8 performs Hasselman inverse transformation on the actual measurement image read from the actual measurement image storage unit 7, and the actual measurement image sea wave Get the spectrum. The acquired actual measurement image sea surface wave spectrum is stored in the sea surface wave spectrum storage unit 9 together with date and time information and position information. Details of the Hasselman inverse transform are disclosed in Non-Patent Document 3 and the like, and thus the description thereof is omitted.
In the second embodiment, as shown in FIG. 7, the user is provided with information on the date when the verification target image was taken and the rough shooting location, and the candidate area L and the candidate time zone T are determined from the provided information. It is set to estimate the shooting time of the verification target image.

  The evaluation comparison unit 6 ′ includes a similarity calculation unit 61 ′, a similarity table storage unit 62, and an estimated time calculation unit 63. The similarity calculation unit 61 ′ calculates the similarity between the verification-target sea surface wave spectrum input from the FFT processing unit 4 and a plurality of actually measured image sea surface wave spectra stored in the sea surface wave spectrum storage unit 9. A similarity table in which the photographing time and the calculated similarity are associated with each other is created and stored in the similarity table storage unit 62. The process by which the estimated time calculation unit 63 estimates the photographing time to be verified based on the similarity table is the same as that in the first embodiment, and a method for obtaining a correlation for each two-dimensional pixel, and other general methods Is applicable.

Next, the operation of the image processing apparatus 10 ′ according to the second embodiment will be described.
FIG. 8 is a flowchart showing the operation of the image processing apparatus according to the second embodiment. The same steps as those of the image processing apparatus 10 according to the first embodiment are denoted by the same reference numerals as those used in FIG. 2, and the description thereof is omitted or simplified.
The FFT processing unit 4 performs fast Fourier transform on the three-dimensional image of the sea surface wave generated in step ST4, and obtains a verification target sea surface wave spectrum (step ST5 ').
On the other hand, when the verification target image is read in step ST1, the Hasselman inverse transform unit 8 acquires a measured image satisfying the candidate region L and the candidate time zone T from the measured image storage unit 7 (step ST31). The actually measured image is subjected to Hasselman inverse transformation to acquire the actually measured image sea surface wave spectrum and stored in the sea surface wave spectrum storage unit 9 together with the date and time information and the position information (step ST32).

  The evaluation comparison unit 6 ′ estimates the time when the verification target image was captured by comparing the actually measured image sea surface wave spectrum stored in step ST31 with the verification target image sea surface wave spectrum acquired in step ST5 ′. Step ST33), the estimated time is output as estimated date and time information (step ST8), and the process is terminated.

Next, the operation of the Hasselman inverse transform unit 8 will be described.
FIG. 9 is a flowchart illustrating the operation of the Hasselman inverse transform unit of the image processing apparatus according to the second embodiment.
The Hasselman inverse transform unit 8 reads the measured image at time t and position c that satisfy the conditions of the candidate region L and the candidate time zone T from the measured image storage unit 7 (step ST41). The actual measurement image read out in step ST41 is subjected to Hasselman inverse transform to acquire the actual measurement image sea surface wave spectrum (step ST42). Store (step ST43). Thereafter, it is determined whether or not the measured image sea surface wave spectrum has been acquired for the measured images at all times in the candidate time zone T (step ST44). When the measured image sea surface wave spectrum is not acquired for the measured images at all times (step ST44; NO), the time interval τ is added to the time t (step ST45), and the process returns to step ST41. On the other hand, when the measured image sea surface wave spectrum is acquired for the measured images at all times (step ST44; YES), the process is terminated.

Next, the operation of the similarity calculation unit 61 ′ of the evaluation comparison unit 6 ′ will be described.
FIG. 10 is a flowchart illustrating the operation of the similarity calculation unit of the image processing apparatus according to the second embodiment. Note that the same steps as those in the similarity calculation unit 61 of the first embodiment are denoted by the same reference numerals as those used in FIG. 4, and the description thereof is omitted or simplified.
When the verification target sea surface wave spectrum is input from the FFT processing unit 4 (step ST51), the similarity calculating unit 61 ′ obtains an actually measured image sea surface wave spectrum at time t and position c from the sea surface wave spectrum storage unit 9 ( Step ST52). Next, the similarity between the verification-target sea surface wave spectrum acquired in step ST51 and the actually measured image sea surface wave spectrum acquired in step ST52 is calculated (step ST53). And are written in the similarity table (step ST14). Thereafter, it is determined whether or not the similarity is calculated for all times t in the candidate time zone T (step ST15). When the similarity is not calculated for all times t (step ST15; NO), the time interval τ is added to the time t (step ST16), and the process returns to step ST52. On the other hand, when the similarity is calculated for all times t (step ST15; YES), the similarity table is stored in the similarity table storage unit 62 (step ST17), and the process is terminated.

  Since the processing operation of the estimated time calculation unit 63 using the similarity table generated by the similarity calculation unit 61 ′ described above is the same as that in the first embodiment, description thereof is omitted.

  As described above, according to the second embodiment, the scale of the verification target image is estimated from the shape of the moving body of the moving body image, and a three-dimensional image of the sea surface wave is generated from the background image based on the estimated scale. Scale estimation unit 3, FFT processing unit 4 that acquires the sea surface wave spectrum, Hasselman inverse transform processing that performs the Hasselman inverse transform process on the measured image and acquires the measured image sea surface wave spectrum, and the sea surface wave spectrum to be verified And the evaluation image sea surface wave spectrum, and the evaluation comparison unit 6 ′ for estimating the time at which the verification target image was captured is provided. Therefore, the verification target image is obtained from the sea wave of the moving body image and the background image. Can be estimated. Specifically, the sea surface wave spectrum of the measured image with the known shooting time and the sea surface wave spectrum of the verification target with the unknown shooting time are compared with each other based on the moving object in the verification target image, and the verification target image is shot. Time can be estimated.

  Further, according to the second embodiment, the similarity indicating the similarity between the verification target sea surface wave spectrum and the actually measured image sea surface wave spectrum is calculated, and the calculated similarity and the photographing time of the actually measured image are shown in association with each other. The similarity calculation unit 61 ′ that creates the similarity table, the time t1 at which the degree of similarity becomes maximum, and the time t2 at which the degree of similarity becomes the second largest are selected, and the verification target image is based on the selected times t1 and t2. The estimated time calculation unit 63 that estimates the shooting time is provided, so that the time when the verification target image was shot can be estimated based on the similarity of the images.

  As described above, according to the second embodiment, when estimating the time when a sea surface-like moving body is captured using the sea surface wave of the background image, the verification target sea surface wave spectrum uses the moving object such as a ship and the background. The correct shooting time can be calculated even when the mutual position of the image appears to deviate from the actual position. As a result, it is judged whether or not the accident situation photo and the report information (time, place, etc.) submitted by the ship are correct, as well as a judgment document for ship accidents, marine accident trials, compensation for environmental damage, etc. or reference materials. be able to.

  In the above-described second embodiment, the configuration in which information related to the candidate region L and the candidate time zone T is input to the Hasselman inverse transform unit 8 from the outside has been described. It may be configured such that the Hasselman inverse transform is performed on all actually measured images stored in the storage unit 7 to obtain the actually measured image sea surface wave spectrum, and the similarity calculating unit 61 ′ calculates the similarity. In addition, when the verification target image is input as electronic data, the candidate time zone T may be set from the shooting time included in the electronic data. Thus, the setting method of the candidate area | region L and the candidate time slot | zone T can be changed suitably.

  In the first embodiment and the second embodiment described above, an example in which one position c is assigned to the candidate area L has been described. However, a plurality of positions c1, c2,. It is also possible to calculate the similarity for the positions c1, c2,... Cn and time t.

  In the present invention, within the scope of the invention, any combination of the embodiments, or any modification of any component in each embodiment, or omission of any component in each embodiment is possible. .

  DESCRIPTION OF SYMBOLS 1 Reading part, 2 Separation part, 3 Scale estimation part, 4 FFT processing part, 5 Hasselman conversion part, 6,6 'Evaluation comparison part, 7 Actual measurement image storage part, 8 Hasselman reverse conversion part, 9 Sea surface wave spectrum storage Sections, 10, 10 ′ image processing apparatus, 61, 61 ′ similarity calculation section, 62 similarity table storage section, 63 estimated time calculation section.

Claims (4)

A separation unit that separates verification target image data obtained by photographing a moving body on the sea surface into a moving body image indicating the moving body and a background image including sea waves other than the moving body;
A scale estimation unit that estimates the three-dimensional coordinates of the sea surface wave of the background image based on the scale of the moving body in the moving body image separated by the separation unit;
An FFT processing unit for performing a fast Fourier transform on the three-dimensional coordinates of the sea surface wave estimated by the scale estimation unit and obtaining a sea surface wave spectrum;
A Hasselman transform that obtains an estimated synthetic aperture radar image by Hasselman transforming the sea surface wave spectrum obtained by the FFT processor;
An actual image storage unit for storing a plurality of actual images obtained by imaging a predetermined area at a preset time interval together with an imaging time;
A similarity indicating the similarity between the estimated synthetic aperture radar image acquired by the Hasselman conversion unit and each measured image stored in the measured image storage unit is calculated, and the photographing time of the measured image indicating a high similarity is calculated. An image processing apparatus comprising: an evaluation comparison unit that estimates a photographing time of the verification target image data. The evaluation comparison unit
The similarity between the estimated synthetic aperture radar image acquired by the Hasselman transform unit and each measured image stored in the measured image storage unit is calculated, and the photographing time of the measured image corresponding to the calculated similarity is calculated. A similarity calculation unit that creates a similarity table showing
A similarity table storage unit that stores a similarity table created by the similarity calculation unit;
When the similarity table stored in the similarity table storage unit is referenced, the top two values having the highest similarity are selected, and the shooting times of the two actually measured images corresponding to the selected binary similarity are consecutive. The selected binary similarity is weighted, and the shooting time at which the similarity is maximized is estimated as the shooting time of the verification target image data, and 2 corresponding to the selected binary similarity. The image processing apparatus according to claim 1, further comprising: an evaluation comparison unit that estimates a shooting time of the two actually measured images as a shooting time of the verification target image data when the shooting times of the two actually measured images are not continuous. A separation unit that separates verification target image data obtained by photographing a moving body on the sea surface into a moving body image indicating the moving body and a background image including sea waves other than the moving body;
A scale estimation unit that estimates the three-dimensional coordinates of the sea surface wave of the background image based on the scale of the moving body in the moving body image separated by the separation unit;
An FFT processing unit that performs fast Fourier transform on the three-dimensional coordinates of the sea surface wave estimated by the scale estimation unit, and obtains a verification target image sea surface wave spectrum;
An actual image storage unit for storing a plurality of actual images obtained by imaging a predetermined area at a preset time interval together with an imaging time;
Hasselman inverse transform of the actual measurement image stored in the actual measurement image storage unit to obtain the actual measurement image sea surface wave spectrum; and
The sea surface wave spectrum storage unit that stores the measured image sea surface wave spectrum acquired by the Hasselman inverse transform unit together with the photographing time;
Calculate the similarity indicating the similarity between the verification target image sea surface wave spectrum acquired by the FFT processing unit and the actual image sea surface wave spectrum stored in the sea surface wave spectrum storage unit, and measure the actual image sea surface indicating the high similarity An image processing apparatus comprising: an evaluation comparison unit configured to estimate a shooting time of the verification target image data using a shooting time of an actually measured image having a wave spectrum. The evaluation comparison unit
The similarity between the verification target image sea surface wave spectrum acquired by the FFT processing unit and the actually measured image sea surface wave spectrum stored in the sea surface wave spectrum storage unit is calculated, and the measured image corresponding to the calculated similarity is calculated. A similarity calculation unit that creates a similarity table showing the shooting time in association with each other;
A similarity table storage unit that stores a similarity table created by the similarity calculation unit;
When the similarity table stored in the similarity table storage unit is referenced, the top two values having the highest similarity are selected, and the shooting times of the two actually measured images corresponding to the selected binary similarity are consecutive. The selected binary similarity is weighted, and the shooting time at which the similarity is maximized is estimated as the shooting time of the verification target image data, and 2 corresponding to the selected binary similarity. The image processing apparatus according to claim 3, further comprising: an evaluation comparison unit that estimates a shooting time of the two actually measured images as a shooting time of the verification target image data when the shooting times of the two actually measured images are not consecutive.

Images