JP4168040B2 - Embedded object search processing method and apparatus, embedded object search processing program, and recording medium recording the program - Google Patents

Description

  The present invention relates to a buried object exploration processing method and apparatus for searching for buried objects in exploration objects such as concrete and underground, a buried object exploration processing program using the buried object exploration processing method, and The present invention relates to a recording medium on which a program is recorded.

  In recent years, undergrounding of gas, water supply, communication / power cables, etc. has been promoted, but there are not enough drawings to accurately record the position of such existing buried pipes, so there is a damage accident in road construction etc. Has occurred. In addition, in the construction of concrete slabs and walls through the installation of electrical equipment, information and communication equipment, air conditioning and sanitary equipment for existing buildings accompanying the increase in the number of renovation starts, water pipes, gas pipes, and electric pipes during construction The number of accidents such as damage is increasing. For this reason, in order to prevent the above-mentioned accidents, it is necessary to know the internal structure of the surroundings where the work is to be performed in advance. It's being used.

  For example, in the electromagnetic wave method, an electromagnetic wave is irradiated from the transmitting antenna to the ground or a structure and propagated, and the reflected wave when it hits an object with different electrical constants such as dielectric constant is measured by the receiving antenna. Since the reflected signal is radiated at an angle and the antenna is moved to receive the reflected signal, the reflected image of the embedded tube is a hyperbola in the cross-sectional image, and the signal of other embedded objects is also superimposed, so that the received signal is the detection target Therefore, it is very difficult to distinguish between the pipes from the buried pipe and those other than the above, and there is a problem that skilled knowledge is required and that the operator needs to be a skilled person.

  In addition, in the ultrasonic method, ultrasonic waves are reflected regardless of metal or non-metal, it is possible to explore deep buried positions, there are no adverse effects on the human body, and the features such as easy handling, It is attracting attention as the next generation non-destructive inspection technology. However, the non-destructive inspection using ultrasonic waves is widely used as a flaw detection method (crack exploration) of a specimen, but is not so much applied to estimation of a pipe configuration in concrete. This is due to the fact that the concrete is scattered and attenuated due to the porosity, which is not constant due to the inhomogeneity inside the concrete, making it difficult to analyze the waveform. As described above, since scattering and attenuation are not constant inside the concrete, there is a problem that it is very difficult to compare the amplitude values of the received waveforms.

  In order to solve the above problems, the present inventor disclosed in Patent Document 1 “Searching for an electromagnetic wave or ultrasonic transmission wave through a buried object continuously extending in a predetermined length direction in an exploration object. In the buried object exploration processing method for radiating from the surface of the object, receiving the reflected wave of the transmission wave, and exploring based on the reflected wave data, the exploration of the buried object is performed based on the received reflected wave data. Extracting data of predetermined feature parameters for estimating the position of the embedded object using knowledge that the embedded object extends continuously based on the extracted feature parameter data And a step of generating and outputting an image of the embedded object in the form of a three-dimensional image based on the estimated position of the embedded object.

  Specifically, the buried object exploration system using this buried object exploration processing method scans two directions of the X axis direction and the Y axis direction orthogonal to each other using a pulse radar sensor at a constant interval, and generates a received amplitude. As a result of obtaining a tomographic image of 256 gradations, constructing input data from a plurality of scanned images, and applying an analysis algorithm, output data is expressed in three dimensions. The buried pipe position is obtained by importing the buried pipe data obtained from the input image into the system and applying an extraction algorithm to measure each parameter such as the measurement parameters related to the expert's knowledge such as density and hyperbolic shape, the buried depth, and the surrounding conditions. It is estimated by performing a calculation belonging to the buried pipe from the measurement parameters related to the priority of knowledge. Specifically, these measurement parameters include parameters related to concentration knowledge, parameters related to hyperbolic shape, parameters related to the estimated position of the buried pipe, etc., and these measurement parameters are empirically set in advance based on past experiments. Has been.

JP 2000-338255 A. JP 2000-212267 A. JP 2001-311781 A. Japanese Patent Application Laid-Open No. 11-006879. JP-A-8-152481. JP-A-9-281229.

  The conventional buried pipe estimation process implemented in this buried object exploration system is a process in which the estimation of the buried pipe is suppressed, and in addition, this system has parameters fixed to preferred values in the same exploration environment. Therefore, there has been a problem that the phenomenon of unextracted buried pipes appears depending on the exploration environment. In that case, it is necessary to manually adjust and search for unextracted buried pipes, but since there are multiple measurement parameters and their combinations are enormous, the measurement parameters cannot be set to the optimum values, and the conventional accuracy May not be obtained. Further, the relationship between the parameter and the number of buried pipes extracted is non-linear, and there is a problem that the optimum value of the measurement parameter cannot be easily determined.

  The object of the present invention is to solve the above-mentioned problems and to execute the buried object searching process while adaptively controlling the measurement parameter used when executing the buried object searching process to the optimum value. A buried object exploration processing method and apparatus, a buried object exploration processing program, and a recording medium in which the program is recorded, which enables an easy and high-accuracy search for a buried object and allows anyone to easily confirm the position of the buried object by visual recognition. Is to provide.

The embedded object exploration processing method according to the present invention radiates an electromagnetic wave or an ultrasonic transmission wave from the surface of an exploration object to an embedded object continuously extending in a predetermined length direction in the exploration object, and transmits the transmission wave. In the buried object exploration processing method for receiving the reflected wave and exploring based on the reflected wave data,
Extracting data of predetermined characteristic parameters for exploring the buried object based on the received reflected wave data;
Based on the extracted feature parameter data, using a genetic algorithm, the embedded object is continuously controlled while adaptively controlling the position estimation processing parameters so that the degree of fitness of the position estimation processing parameters is high. Estimating the position of the buried object using the knowledge of extending;
And generating and outputting an image of the embedded object in the form of a three-dimensional image based on the estimated position of the embedded object.

The buried object exploration processing method according to the present invention radiates an electromagnetic wave or ultrasonic transmission wave from the surface of the exploration object to the buried object continuously extending in a predetermined length direction in the exploration object, In the buried object exploration processing method for receiving the reflected wave of the transmitted wave and exploring based on the reflected wave data,
Based on the received reflected wave data, image data having density data of at least one cross section orthogonal to the surface of the exploration object is generated, and using the knowledge that the embedded object extends continuously. Generating candidate point image data of at least one cross section indicating candidate points of the embedded object by removing image data in which no embedded object exists from the generated image data;
Storing the generated candidate point image data of at least one cross section in an image memory so as to be virtually arranged in a three-dimensional space;
Based on the candidate point image data stored in the image memory, using a genetic algorithm, the embedded object is controlled while adaptively controlling the position estimation processing parameters so that the degree of fitness of the position estimation processing parameters is high. Estimating the position of the buried object by connecting candidate points using the knowledge of extending continuously;
And generating and outputting an image of the embedded object in the form of a three-dimensional image based on the estimated position of the embedded object.

In the buried object exploration processing method, preferably, after the step of estimating the position of the buried object,
(A) the first rule that both ends of the buried object are close to the end of the exploration object, and the depth of the buried object is not between the cover thicknesses of the exploration object layer, the probability of being an embedded object is high;
(B) a second rule that the probability that the object is a buried object is low when both ends of the buried object are far from the end of the exploration object and the depth of the buried object is between the cover thicknesses of the exploration object layer. Using fuzzy rules,
Based on the estimated position of the embedded object, it is determined whether or not the estimated position of the embedded object is an embedded object, and when the estimated position of the embedded object is determined to be not an embedded object, The method further includes a step of removing from the data.

In the buried object exploration processing method, the step of estimating the position of the buried object is preferably,
The knowledge of the density in the candidate point image data stored in the image memory is expressed, and a predetermined first membership function is used to indicate the degree of the buried position candidate based on the average angle of density change. Calculating the degree,
Represents knowledge of hyperbolic shape in candidate point image data stored in the image memory, and indicates the degree of the embedded position candidate based on the degree of coincidence between the hyperbola obtained by referring to the density in the candidate point image data and the theoretical hyperbola. Calculating a degree of the second embedded object using a predetermined second membership function shown;
The degree of the third buried object is calculated by integrating the calculated degree of the first buried object and the calculated degree of the second buried object by linear combination using a function of a predetermined weighting coefficient. Steps,
And a step of estimating the position of the embedded object by determining whether or not the image data is an embedded object candidate based on the degree of the third embedded object.

  Further, in the embedded object exploration processing method, the position estimation processing parameters are preferably a parameter for defining the first membership function, and a parameter for defining the second membership function. And a parameter for defining a function of the weighting coefficient.

  Still further, in the buried object exploration processing method, the fitness is preferably expressed as a sum of lengths of the buried pipes estimated and extracted from the candidate point image data stored in the image memory. Features.

An embedded object exploration processing apparatus according to the present invention radiates an electromagnetic wave or an ultrasonic transmission wave from a surface of an exploration object to an embedded object continuously extending in a predetermined length direction in the exploration object, and transmits the transmission wave. In the buried object exploration processing device that receives the reflected wave of and searches based on the reflected wave data,
Based on the received reflected wave data, extracting means for extracting data of predetermined feature parameters for exploring the buried object;
Based on the extracted feature parameter data, using a genetic algorithm, the embedded object is continuously controlled while adaptively controlling the position estimation processing parameters so that the degree of fitness of the position estimation processing parameters is high. An estimation means for estimating the position of the buried object using the knowledge that it extends,
Output means for generating and outputting an image of the embedded object in the form of a three-dimensional image based on the estimated position of the embedded object.

Further, the buried object exploration processing apparatus according to the present invention radiates electromagnetic waves or ultrasonic transmission waves from the surface of the exploration object to the buried object that continuously extends in the predetermined length direction in the exploration object, and In the buried object exploration processing device that receives the reflected wave of the transmitted wave and searches based on the reflected wave data,
Based on the received reflected wave data, image data having density data of at least one cross section orthogonal to the surface of the exploration object is generated, and using the knowledge that the embedded object extends continuously. Generating means for generating candidate point image data of at least one cross section indicating candidate points of the embedded object by removing image data in which no embedded object exists from the generated image data;
Storing means for storing the generated candidate point image data of at least one cross section in an image memory so as to be virtually arranged in a three-dimensional space;
Based on the candidate point image data stored in the image memory, using a genetic algorithm, the embedded object is controlled while adaptively controlling the position estimation processing parameters so that the degree of fitness of the position estimation processing parameters is high. Position estimation means for estimating the position of the buried object by connecting candidate points using the knowledge of extending continuously;
Output means for generating and outputting an image of the embedded object in the form of a three-dimensional image based on the estimated position of the embedded object.

In the buried object exploration processing apparatus, preferably, after the processing of the position estimation means,
(A) the first rule that both ends of the buried object are close to the end of the exploration object, and the depth of the buried object is not between the cover thicknesses of the exploration object layer, the probability of being an embedded object is high;
(B) a second rule that the probability that the object is a buried object is low when both ends of the buried object are far from the end of the exploration object and the depth of the buried object is between the cover thicknesses of the exploration object layer. Using fuzzy rules,
Based on the estimated position of the embedded object, it is determined whether or not the estimated position of the embedded object is an embedded object, and when the estimated position of the embedded object is determined to be not an embedded object, It is further characterized by further comprising a removing means for removing from the data.

In the buried object exploration processing apparatus, the position estimation means is preferably
The knowledge of the density in the candidate point image data stored in the image memory is expressed, and a predetermined first membership function is used to indicate the degree of the buried position candidate based on the average angle of density change. A first calculating means for calculating the degree;
Represents knowledge of hyperbolic shape in candidate point image data stored in the image memory, and indicates the degree of the embedded position candidate based on the degree of coincidence between the hyperbola obtained by referring to the density in the candidate point image data and the theoretical hyperbola. Second calculating means for calculating a degree of the second embedded object using a predetermined second membership function shown;
The degree of the third buried object is calculated by integrating the calculated degree of the first buried object and the calculated degree of the second buried object by linear combination using a function of a predetermined weighting coefficient. A third calculating means;
And a buried object position estimating means for estimating the position of the buried object by determining whether the image data is a candidate for a buried object based on the degree of the third buried object.

  Furthermore, in the buried object exploration processing device, the position estimation processing parameters are preferably a parameter for defining the first membership function, and a parameter for defining the second membership function. And a parameter for defining a function of the weighting coefficient.

  Still further, in the embedded object exploration processing apparatus, the fitness is preferably expressed as a sum of the lengths of the respective buried pipes estimated and extracted from the candidate point image data stored in the image memory. Features.

  The buried object search processing program according to the present invention includes each step of the buried object search processing method.

  The recording medium according to the present invention is characterized in that the above-described buried object search processing program is recorded.

  Therefore, according to the buried object exploration processing method or apparatus according to the present invention, the buried object continuously extending in the predetermined length direction in the exploration object is transmitted from the surface of the exploration object by electromagnetic waves or ultrasonic waves. In a buried object exploration processing method or apparatus for radiating, receiving a reflected wave of the transmitted wave, and exploring based on the reflected wave data, for exploring the buried object based on the received reflected wave data Based on the extracted feature parameter data, the feature estimation parameter is adaptively controlled based on the extracted feature parameter data using a genetic algorithm so that the fitness of the location estimation process parameter is increased. However, the position of the embedded object is estimated using knowledge that the embedded object extends continuously, and the image of the embedded object is three-dimensionally based on the estimated position of the embedded object. Generating and outputting in the form of an image. Therefore, the embedded object exploration process can be executed while adaptively controlling the measurement parameters used when executing the embedded object exploration process to the optimum values, and the embedded object in various exploration objects can be easily compared with the prior art. Thus, it is possible to search with high accuracy, and anyone can easily confirm the position of the buried object by visual recognition.

  Further, according to the buried object exploration processing method or apparatus according to the present invention, the buried object continuously extending in the predetermined length direction in the exploration object is transmitted from the surface of the exploration object by electromagnetic waves or ultrasonic waves. In a buried object exploration processing method or apparatus for radiating, receiving a reflected wave of the transmitted wave, and exploring based on reflected wave data, orthogonal to the surface of the exploration object based on the received reflected wave data Generating image data having density data of at least one cross section, and using the knowledge that the embedded object extends continuously, removing image data having no embedded object from the generated image data Candidate point image data of at least one cross section indicating a candidate point of the embedded object is generated, and the generated at least one cross section candidate point image data is virtually arranged in a three-dimensional space. The position estimation parameters are adaptively controlled using genetic algorithms based on the candidate image data stored in the image memory and using the genetic algorithm. However, the position of the buried object is estimated by connecting candidate points using knowledge that the buried object continuously extends, and based on the estimated position of the buried object, An image is generated and output in the form of a three-dimensional image. Therefore, the embedded object exploration process can be executed while adaptively controlling the measurement parameters used when executing the embedded object exploration process to the optimum values, and the embedded object in various exploration objects can be easily compared with the prior art. Thus, it is possible to search with high accuracy, and anyone can easily confirm the position of the buried object by visual recognition.

  Embodiments according to the present invention will be described below with reference to the drawings.

<First Embodiment>
FIG. 1 is a flowchart showing a measurement process including an adaptive control process for measurement parameters according to the first embodiment of the present invention. In the first embodiment, in a predetermined measurement process executed using a predetermined measurement parameter, the measurement process is executed while adaptively controlling the measurement parameter. Here, the measurement process is executed by a computer such as a digital computer, for example, and can be configured as a measurement apparatus that executes the measurement process. In addition to the buried object exploration process shown in the second embodiment, the measurement process is widely applied to a measurement process related to a physical phenomenon, a measurement process related to an electrical system or a power distribution system, a measurement process related to a telecommunications system, and the like. can do.

  In FIG. 1, first, input data measured in step S101 is received, and initial values of measurement parameters are generated and set in step S102. Next, based on the input data measured in step S103, a predetermined measurement process is executed using the set measurement parameter to obtain a measurement process result. Then, in step S104, using the genetic algorithm based on the measurement processing result, the measurement parameter is adaptively controlled and set so that the degree of fitness of the measurement parameter is high. In step S105, whether a predetermined termination condition is satisfied. When the end condition is not satisfied, the process returns to step S103, and the above process is repeated. On the other hand, when the end condition is satisfied, the measurement process result in the last step S103 is output as the output data in step S106. Output, and the measurement process ends.

  Here, the termination condition may be terminated if the number of repetitions of the processing from step S103 to step S104 is counted to be a predetermined number, or the difference in parameter values before and after the adaptive control of the measurement parameter. The process may be terminated on the assumption that becomes smaller than a predetermined value and has converged. In addition, the degree of conformity is determined in advance so that the measurement accuracy or accuracy of the measurement process increases as the conformity increases, while the measurement accuracy or accuracy of the measurement process decreases as the conformance decreases. It is a defined scale.

  Each step of the measurement process of FIG. 1 is generated as a program, and the program is stored in a storage device of the computer and then executed by the computer. Alternatively, each step of the measurement process of FIG. 1 is generated as a program, and the program is recorded on a computer-readable recording medium such as a CD-ROM, CD-R, or DVD, and the recorded program is a storage device of the computer. After being loaded, it is executed by the computer.

  As described above, according to the first embodiment, the measurement process can be executed while adaptively controlling the measurement parameter used when executing the predetermined measurement process to the optimum value, and the measurement process is adapted. It can be executed with high accuracy so as to increase the degree.

<Second Embodiment>
FIG. 2 is a block diagram showing the configuration of the buried object search system according to the second embodiment of the present invention, and FIG. 3 shows the buried object search image processing executed by the CPU 20 of the image processing apparatus 10 of FIG. It is a flowchart which shows (main routine).

  The buried object exploration system of this embodiment radiates an electromagnetic wave transmitted from the surface of the exploration object and receives the reflected wave of the transmitted wave from the surface of the exploration object that is continuously present in the exploration object in a predetermined length direction. The exploration is based on the reflected wave data. Here, the image processing apparatus 10 can easily and highly accurately search by executing the embedded object exploration image processing shown in FIG. 3, and in the form of a three-dimensional image showing the position of the embedded object, However, it is characterized in that the position of the buried object can be easily confirmed by visual recognition. In the present embodiment, the buried object is, for example, a pipe shape that is continuously embedded in a predetermined direction, such as a steel pipe, a CD pipe (cable pipe), a copper pipe, and a reinforcing bar, which is buried in concrete or the ground. It is intended for electric pipes, gas pipes, water pipes, etc., which are buried pipes. In the exploration object, the buried pipe may extend continuously in a straight line, or may be bent in a curved line and continuously extended.

As shown in FIG. 2, the image processing system of this embodiment is roughly divided into
(A) After radiating a transmission wave signal of an electromagnetic wave for exploring an embedded object from the surface of the exploration object, receiving the reflected wave signal of the transmission wave and transmitting the data to the image processing apparatus 10 An electromagnetic wave transmitting / receiving device (hereinafter referred to as an electromagnetic wave transmitting / receiving device) 1;
(B) It is configured by a digital computer, and by executing the embedded object search image processing shown in FIG. 3 based on the received reflected wave signal data, the embedded object can be searched easily and with high accuracy. It is displayed or printed in the form of a three-dimensional image showing the position of the object and output.

  As shown in FIG. 23, the electromagnetic wave transmitting / receiving apparatus 1 of FIG. 2 is mounted on a vehicle body 1RR having wheels R1 that are driven and rotated by a motor (not shown) or manually, and a search surface 92 that is the surface of the search object 91. As shown in FIG. 16, the object to be scanned is moved by predetermined intervals Δx and Δy in two directions orthogonal to each other as shown in FIG. The direction orthogonal to the exploration surface 92 is taken as the Z direction, and the distance from the exploration surface 92 is defined as the embedding depth. The electromagnetic wave transmission / reception apparatus 1 includes a transmitter 4 that generates and amplifies a transmission wave signal of an electromagnetic wave having a predetermined frequency spectrum and transmits the signal to a search object via the transmission antenna 2 and the search object of the transmission wave signal. The receiver 5 receives and amplifies the reflected wave signal from the receiver 5 using the receiving antenna 3, samples the reflected signal at a predetermined time timing, and outputs it as reflected wave data. The electromagnetic wave transmitter / receiver 1 also generates and outputs a timing signal indicating the transmission timing of the transmitter 4 and the reception timing of the receiver 5 to control the operations of the transmitter 4 and the receiver 5. And a communication interface 7 that executes signal processing such as signal conversion. The data of the reflected wave signal output from the receiver 5 is transmitted to the image processing apparatus 10 via the communication interface 7 and the communication cable 51. Here, these communication interfaces 7 and 61 are communication interfaces having a predetermined communication protocol and signal format, for example.

Next, the configuration of the image processing apparatus 10 will be described with reference to FIG. The image processing apparatus 10
(A) a CPU (central processing unit) 20 of a computer that calculates and controls the operation and processing of the image processing apparatus 10;
(B) a ROM (read only memory) 21 for storing a basic program such as an operation program and data necessary for executing the basic program;
(C) A RAM (random access memory) 22 that operates as a working memory of the CPU 20 and temporarily stores parameters and data necessary for image processing;
(D) a reception memory 26 configured by, for example, a hard disk memory and storing reflected wave signal data received from the electromagnetic wave transmission / reception device 1;
(E) a processing memory 25 configured by, for example, a hard disk memory and temporarily storing data during processing when executing the embedded object exploration image processing of FIG. 3;
(F) An image memory 23 configured, for example, by a hard disk memory and temporarily storing image data when executing the embedded object exploration image processing of FIG.
(G) a program memory 24 configured by, for example, a hard disk memory and storing the embedded object search image processing program of FIG. 3 read using the CD-ROM drive device 45;
(H) a communication interface 61 that is connected to the communication interface 7 of the electromagnetic wave transmission / reception device 1 and transmits / receives data to / from the communication interface 7;
(I) A keyboard that is connected to a keyboard 41 for inputting predetermined data and instruction commands, receives the data and instruction commands input from the keyboard 41, performs interface processing such as predetermined signal conversion, and transmits the data to the CPU 20 Interface 31;
(J) Connected to a mouse 42 for inputting instruction commands on the CRT display 43, receives data and instruction commands input from the mouse 42, performs predetermined signal conversion and other interface processing, and transmits them to the CPU 20. A mouse interface 32;
(K) Connected to a CRT display 43 that displays image data processed by the CPU 20, a setting instruction screen, and the like, converts image data to be displayed into an image signal for the CRT display 43, and outputs the image signal to the CRT display 43 for display. Display interface 33,
(L) A printer interface 34 that is connected to a printer 44 that prints image data processed by the CPU 20, predetermined analysis results, and the like, performs predetermined signal conversion of print data to be printed, and outputs the data to the printer 44 for printing. When,
(M) Connected to a CD-ROM drive device 45 that reads program data of an image processing program from a CD-ROM 45a in which an embedded object search image processing program is stored, and the read program data of the embedded object search image processing program is predetermined. And a drive device interface 35 for performing signal conversion and transferring to the program memory 24, and these circuits 20-26, 31-35 and 61 are connected via a bus 30.

  Next, the measurement method will be described. In the present embodiment, the search is performed using an electromagnetic wave including a broadband frequency component. FIG. 14 shows the frequency spectrum of the electromagnetic wave transmitted from the transmitter 4. The pulse width at the output end of the pulse generator in the transmitter 4 is about 0.7 nsec. Further, when output from the transmission antenna 2 as a radio wave, the band is limited by the transmission antenna 2 so that the pulse width changes somewhat, but the emitted electromagnetic wave has a waveform as shown in FIG.

  In the present embodiment, as shown in FIG. 16, an exploration device including the electromagnetic wave transmission / reception device 1 having the transmitting antenna 2 and the receiving antenna 4 on the surface of the test body that is the exploration object 91 is arranged in the X and Y directions orthogonal to each other. , And continuously explore at constant intervals Δx and Δy. Electromagnetic waves are driven into the probe object 91 from the transmission antenna 2 at intervals of 5 mm. As shown in FIG. 17, the electromagnetic wave is reflected at a location where the electrical property is different from that of the surrounding medium, such as a buried pipe, and therefore the distance to the buried object is measured from its propagation time. The distance to the buried object is obtained by the following formula. In this specification, the number number of the black brackets in which the mathematical formula is imaged and the formula number of the square brackets in which the mathematical formula is input are used in combination. The formula number is assigned to the last part of the formula using the format of “formula (1)” as the formula number.

[Equation 1]
V = C / √ (ε) [m / sec] (1)
[Equation 2]
D = (VT) / 2 [m] (2)

  Where V is the velocity of electromagnetic waves in the concrete, C is the velocity of electromagnetic waves in vacuum, ε is the relative permittivity of concrete, D is the distance to the buried object, and T is the electromagnetic wave Is the round-trip propagation time.

  In addition, since this exploration device does not have perfect directivity and receives a wide range of reflected waves, even if there is no buried object directly under the exploration point, it receives the reflected wave if it exists in the vicinity. . However, since the propagation time is different from that in the case where the buried object exists immediately below, the reflected wave group of the buried object has a hyperbolic shape as shown in FIG. As shown in FIG. 19A, the received wave is composed of a reflection component on the surface, a reflection component of the buried object, and a reflection component (noise waveform) of the buried object in a wide area. Then, based on the amplitude of the received wave, the search result is expressed as a grayscale image with 256 gradations as shown in FIGS. 19B and 19C, for example.

  Next, characteristics of the received waveform will be described. The reflected wave group of the buried object basically has a characteristic shape called a hyperbola. However, this shape is when the burial depth is shallow or when a single burial object is explored, and when there are multiple burial depths and there are multiple burial objects, the influence of mutual interference of reflected waves is greatly affected. However, the reflected wave group does not necessarily form a hyperbolic pattern.

  In the present embodiment, attention is paid to the amplitude of the received wave. The received waveform and the search result image when there is an embedded object are shown in FIGS. 20 and 21, respectively. The waveform shown in FIG. 20B is a waveform observed on the La line in FIG. 20A, and is an example in which a reflected wave of a basic buried object is received. As shown in FIG. 20, since the reflected wave of the embedded object has a clearly large peak, the embedded position can be easily determined. On the other hand, the waveform shown in FIG. 21B is a waveform observed on the Lb line in FIG. In this waveform, mutual interference occurs between the reflected wave of the buried object and the reflected wave of the surface of the test object, and a large distortion occurs in the received waveform. This distortion has a large effect on the reflected wave of the buried object having a shallow depth, and suppresses the increase in amplitude as shown in FIG. In this case, it is difficult to find a clear difference from the noise waveform, and as a result, it is difficult to determine the embedded position. However, the interference wave suppresses the reflected wave mainly due to an increase in amplitude, and the decrease in amplitude is not suppressed as much as the increase in amplitude, and the characteristic is that it drops to a sufficiently small value compared to the noise waveform. Since the portion where the amplitude decreases is a dark portion on the image, the embedded position can be estimated by limiting the target to the low density region. From the above characteristics, in the present embodiment, the embedded position is estimated by paying attention to the hyperbolic pattern of the low concentration region and the reflected wave group.

  Next, the embedded object exploration image processing (main routine) will be described with reference to FIG. In FIG. 3, first, in step S1, the reflected wave signal data is received and stored. Here, the data of the reflected wave signal transmitted from the electromagnetic wave transmitting / receiving apparatus 1 is received via the communication interface 61 and stored in the reception memory 26. Next, in step S2, an initial value (predetermined value) of an estimation processing parameter (corresponding to the measurement parameter in the first embodiment) is generated and set, and in step S3, reflection during scanning in the X direction is performed. The embedded position candidate estimation process (FIG. 4) based on the wave signal data is executed. Here, among the reflected wave signal data in the reception memory 26, the hyperbolic pattern is focused on the basis of the reflected wave signal data at the time of scanning in the X direction (hereinafter referred to as XZ slice image data). Data)) is extracted and stored in the image memory 23. Further, in step S4, an embedded position candidate estimation process (FIG. 5) based on the reflected wave signal data at the time of scanning in the Y direction is executed. Here, among the reflected wave signal data in the reception memory 26, the hyperbolic pattern is focused on the basis of the reflected wave signal data at the time of scanning in the Y direction, and image data of the YZ slice of the embedded position candidate (hereinafter referred to as an image of the YZ slice). Data)) is extracted and stored in the image memory 23. In these estimation processes in steps S3 and S4, the original image is quantized into three regions in order to extract a low density region of the image, and then the quantized region is labeled, and the hyperbolic pattern of the reflected wave group and the reflected wave are reflected. The position of the buried object is estimated using the amplitude feature.

In step S5, the data of the two estimation processing results are virtually assigned addresses and stored in the image memory 23 so as to be combined and arranged in the three-dimensional space. Further, after executing the link search process (FIGS. 10 and 11) for generating the link data by connecting and searching the candidate points of the embedded position in step S6, a predetermined fuzzy rule is based on the link data in step S7. The embedded object extraction process (FIG. 12) for determining whether the object is an embedded object such as an embedded pipe continuously extending in a predetermined direction is executed. Next, in step S8, a process (FIG. 13) is executed to adaptively control and set the estimation process parameters using a genetic algorithm based on the extraction process result. In step S9, whether a predetermined end condition is satisfied or not is determined. If the determination is NO, the process returns to step S3 and the above process is repeated. If the determination is YES, the process proceeds to step S10. Here, as an end condition, for example,
(A) Counting the number of iterations in which the processing from step S3 to step S9 has been performed, and exceeding a predetermined iteration number threshold value (a threshold that is in a converged state as will be described later with reference to FIG. 39) Value), or
(B) The difference between the estimation processing parameters before and after the adaptive control processing in step S8 is equal to or less than a predetermined threshold value (a convergence state in which the difference between the estimation processing parameters before and after the adaptive control processing hardly changes). When)
Can be set.

  Finally, in the embedded object image output process in step S10, an image of the embedded object in the exploration object determined to be an embedded object in the embedded object extraction process (an image showing the internal structure of the structure) is a three-dimensional image. Is displayed on the CRT display 43 or printed by the printer 44 and output, and the buried object search image processing is terminated.

  Further, the candidate concentration extraction process performed in steps S3 and S4 will be described in detail. The density distribution of the search result image is shown in FIG. First, in order to extract a low concentration region corresponding to a portion where the amplitude decreases, threshold processing is performed with the average value Th of the amplitude. Next, a known histogram equalization method is applied to a density region that is equal to or lower than the threshold value Th, and is quantized to two gradations. FIG. 23B shows the result of applying this processing to the original image shown in FIG. As a result, three regions are obtained: a low-concentration region A with a high probability that an embedded object exists, a mixed region B of noise and embedded material, and a region C other than A and B. The surface reflection component is previously removed based on the depth information.

  Among the extracted regions, there is a low-concentration region with a high probability that an embedded object exists in the center like the region RRa in FIG. 24, and a region where a mixed region of noise and embedded material exists so as to surround this region. There is a very high probability that it is an image of a buried object. However, it is considered that there is a low probability that an area composed only of a mixed area of noise and embedded objects, such as the area RRb in FIG. 24, is an image of the embedded object. Furthermore, if this area does not have continuity in the preceding and following slices, it can be said that it is not an image of an embedded object. Based on the above knowledge, a region that is composed only of a mixed region of noise and embedded objects and has no continuity is removed. As a result, only a region having a high probability of being an embedded image remains as a candidate region.

  Next, the embedded position estimation process performed in the estimation processes in steps S3 and S4 will be described in detail. Labeling is performed on the embedded candidate region obtained by the previous extraction process. The extraction result shown in FIG. 23B can be divided into seven regions (N1 to N7) as shown in FIG. 25A. Next, the embedded position is estimated with reference to the original image data for each region. The reflected wave of the buried object has a larger amplitude waveform if the distance from the exploration point to the buried object is closer, that is, if the buried object exists immediately below the exploration point. In this embodiment, since attention is paid to the low-concentration region, the minimum point of each candidate region can be obtained, and this point can be used as the estimated position of the embedded object.

  Assuming that all candidate areas are not affected by the mutual interference of reflected waves, one estimated position per area can be basically listed. However, there is a region having three minimum points, such as a region N4 shown in FIG. Since there are two buried pipes in this area and they are close to each other, the reflected waves are strengthened by mutual interference, and an image of the same density area as the image of the buried object is generated in a place where nothing originally exists. As a result, there are three minimum points. Since it is difficult to determine which part is strengthening at this stage, all three places are listed as candidates for burial positions. FIG. 25B shows the estimation result of the buried position by the local minimum point search.

  The influence of the mutual interference of the reflected waves can be cited as well. In the buried object candidate region shown in FIG. 27, since there are a plurality of buried objects near the surface, the influence of the mutual interference attenuates the reflected wave of the buried object, making it difficult to search by the minimum point. In such a region, a buried position can be estimated using a hyperbolic pattern which is a characteristic of the reflected wave group. First, the valleys of candidate areas are extracted with reference to the density of the original image. Next, when the reflected wave group forms a hyperbola pattern, the buried object corresponds to the position of the vertex of the convex hyperbola, so the vertex of the convex hyperbola of the valley line is extracted and used as the estimated position of the buried object.

  Considering the above factors, the embedded position is estimated using both the minimum point and the hyperbolic pattern. In some cases, an estimated position based on a minimum point and an estimated position based on a hyperbolic pattern are obtained in one region. In this case, priority is given to the estimated position by the hyperbolic pattern.

  Next, the estimated position coupling process performed in step S6 will be described in detail. The estimated position of the buried pipe extracted by the above estimation process is a point on the XZ slice or YZ slice, and corresponds to a point on the space in the exploration area. For this reason, the estimation result considered to have estimated the same pipe | tube is connected. First, the extracted total embedded position estimation results in the X direction and the Y direction are arranged in a three-dimensional space. Next, as shown in FIG. 28, a search is performed within a sphere within a predetermined radius from candidate points to be processed based on each estimation result, and the points between the estimated positions are connected to the shortest point. At this time, concatenation is performed with priority given to estimation results of slices in the same direction. That is, in the case of XZ slice image data, the estimation results in the X direction are connected with priority, and the estimation results in the Y direction, which is another direction, are connected only when there is no candidate point. In the case of YZ slice image data, the Y direction estimation results are connected with priority, and the X direction estimation results, which are different directions, are connected only when there is no candidate point.

  Further, the buried pipe extraction process performed in step S7 will be described in detail. In this extraction process, the connection result of the estimated position is evaluated based on the knowledge of the shape, continuity and depth of the buried pipe, and the position of the buried pipe is extracted with higher accuracy by extracting only the buried pipe. . Since the target buried pipes in this embodiment are mainly electric lines, gas pipes, water pipes, etc., the buried pipes are not interrupted in the middle of the exploration area and are connected from end to end within the exploration area. Assuming that Moreover, the concrete structure which is an Example has the cover thickness which is a concrete only layer several cm between the concrete surface and internal piping on the structure, and there is no buried pipe between this cover thickness. Based on the above knowledge, the following rules can be created. Fuzzy reasoning is performed according to this rule, and buried pipes are extracted. FIG. 29 shows a membership function.

[Equation 3]
Rule 1:
“As a result of connecting IF estimated positions, both ends are close to the end of the exploration area,
The depth of the buried pipe is not between the cover thicknesses.
There is a high probability of being a THEN buried pipe (grade A). “(3)
[Equation 4]
Rule 2:
“As a result of connecting IF estimated positions, both ends are far from the end of the exploration area,
The depth of the buried pipe is located between the cover thicknesses.
The probability of being a THEN buried pipe is low (grade B). “(4)

Here, the shortest distance from the end points of the connection result to the outer periphery of the search area is used for the near distance from the search area, and the average value of the depth of the connection result is used for the depth. The degree of G _P of the distance from the outer periphery of the estimated position concatenation, the degree of depth and G _D, to a degree which is buried pipe grade A, the degree not buried pipes and grade B. Grade A and grade B are obtained by the following equations.

[Equation 5]
grade A
_{= W 1 G PNear + W 1} G PNear + W 2 G DMiddle (5)
[Equation 6]
grade B
= W ₁ G _PFar + W ₁ G _PFar + W ₂ G _DDeep + W ₂ G _DShallow (6)

Here, W ₁ and W ₂ are weighting factors, and are set to 1 in the preferred embodiment. Then, the connection result satisfying grade A> grade B is determined as a buried pipe which is a buried object extending continuously. In the present embodiment, for example, the cover thickness is 3 cm, the thickness of the exploration object in the embedment depth direction (Z direction) is 100 cm, the distance from the outer periphery is 3 cm, and the distance is 3 km, and the distance is 6 cm. At this time, each parameter L1, L2, L3, L4, L5 of FIG. 29 is as follows.
L1 = 3 cm.
L2 = 6 cm.
L3 = 3 cm.
L4 = 94 cm.
L5 = 97 cm.

  Further, the processing of each subroutine of FIG. 3 excluding step S8 will be described in detail below.

  FIG. 4 is a flowchart showing an embedded position candidate estimation process (step S3) based on reflected wave signal data at the time of scanning in the X direction, which is a subroutine of FIG. In FIG. 4, first, in step S11, the reflected wave signal data at the time of scanning in the X direction is converted into density data of 256 gradations according to the amplitude, and image data of an XZ slice having density data is generated. Next, in step S12, as shown in FIG. 22, the image data of the XZ slice is quantized into three areas in the light and shade direction, and in step S13, the image data of the low density area is extracted and used as the embedded position candidate image data. Then, in step S14, as shown in FIG. 24, the image data of the region having no continuity is removed from the image data of the XZ slice as the embedment position candidate based on the front-rear relationship in the Y direction. Further, in step S15, as shown in FIG. 25 and FIG. 26, a process for estimating the image data of the embedded position candidate by searching for the minimum point of the shade based on the image data of the XZ slice of the embedded position candidate after the removal (FIG. 6) is executed. Finally, in step S16, image data of the estimated embedment position candidate XZ slice is extracted and stored in the image memory 23, and then the process returns to the original main routine.

  FIG. 5 is a flowchart showing an embedded position candidate estimation process (step S4) based on the reflected wave signal data at the time of scanning in the Y direction, which is a subroutine of FIG. In FIG. 5, first, in step S21, the reflected wave signal data at the time of scanning in the Y direction is converted into density data of 256 gradations according to the magnitude of the amplitude, and image data of YZ slices having density data is generated. Next, in step S22, as shown in FIG. 22, the image data of the YZ slice is quantized into three areas in the light and shade direction, and in step S23, the image data of the low density area is extracted and used as the embedded position candidate image data. Then, in step S24, as shown in FIG. 24, the image data of the region having no continuity is removed from the image data of the YZ slice as the embedment position candidate based on the front-rear relationship in the X direction. Further, in step S25, as shown in FIG. 25 and FIG. 26, a process of estimating the image data of the embedded position candidate by searching for the minimum point of the shade based on the image data of the YZ slice of the embedded position candidate after the removal, The same processing as that in FIG. 6 is executed. Finally, in step S26, the image data of the estimated YZ slice of the embedded position candidate is extracted and stored in the image memory 23, and then the process returns to the original main routine.

  Next, the embedded position estimation method in the process of estimating the embedded position candidate image data in step S15 of FIG. 4 and step S25 of FIG. 5 will be described below.

First, the knowledge about density will be described below. In the grayscale image data obtained up to step S14 in FIG. 4 or step S24 in FIG. 5, the buried pipe appears as a local density change as shown in FIG. For this reason, the search for the concentration minimum point is effective as a means for estimating the buried position. However, since a feature similar to this change in density is also generated by interference of reflected waves as shown in FIG. 30B, it is difficult to estimate the buried position only by searching for a minimum point. Therefore, in the method according to the present embodiment, attention is paid to the density change from the minimum point to the periphery thereof. As shown in FIGS. 30 (a) and 30 (b), the concentration change around the minimum point is an obtuse angle in the vicinity of the buried pipe, whereas it is an acute angle in the interference region. At this time, if the angle of density change is θ _az , the fuzzy If-Then rule shown below can be constructed.

[Equation 7]
If the angle θ _{az of the} concentration change is acute (Acute), the degree of belonging of the Ten buried pipe is low (Low). (7)
[Equation 8]
If the angle θ _{az of} concentration change is obtuse (Obtuse), the degree of belonging of the Ten buried pipe is high (High). (8)
[Equation 9]
Angle theta _az is rectangular of If concentration change (Straight), affiliation of the Then buried pipe is low (Low). (9)

The degree _μgI belonging to the buried pipe based on the knowledge of the concentration is obtained by a known Min-Max centroid method from the membership function shown in FIG. In the input membership function of FIG. 31A, the following values are parameters for estimation processing according to the present embodiment.
(A) Flat angles T1 and T5 of density change when the grade of membership function of “obtuse angle” is zero.
(B) The flat angle T2 of the density change at the boundary when the grade of the membership function of “acute angle” decreases from 1.
(C) Flat angle T3 of density change when the grade of membership function of “acute angle” and “flat angle” becomes zero.
(D) Flat angle T4 of the density change at the boundary when the grade of membership function of “flat angle” is 1.

That is, in the example of FIG. 31, when the average angle θ _{az of} density conversion is T _in , it intersects with the function lines of the “obtuse angle” and “flat angle” membership functions, and these intersection points become the output membership function. extending horizontally, wherein the membership area grades below crossing the function curve of the function a _H of the "high" straight line of the output, extending from the intersection of the "obtuse", a straight line extended from the intersection of the "flat" seeking centroid mu _gI in the area of union of the membership functions of the function curve and intersecting grade following area a _L of the "low" output, that it with the output value mu _gI degree of buried pipe.

Next, knowledge about the hyperbolic shape will be described below. In the grayscale image data obtained up to step S14 in FIG. 4 or step S24 in FIG. 5, the buried pipe appears as a hyperbolic shape shown in FIG. The embedded position can be estimated by searching for this vertex. Here, the hyperbolic shape is obtained as shown in FIG. 32B by binarization processing in which the minimum point of the received wave is “0” and the others are “1”. The degree of coincidence α between the hyperbola obtained from the image data and the theoretical hyperbola (see FIG. 18 or FIG. 32 (c)) is obtained, and belongs to the buried pipe based on the knowledge of the hyperbola shape by the fuzzy If-Then rule shown below. The degree μg _HP is determined. As for the degree of coincidence α between the hyperbola obtained from the image data and the theoretical hyperbola, for example, the maximum point of the theoretical hyperbola is associated with the maximum point of the hyperbola obtained from the image data, and the image data is finely meshed. The number of matching sections in the sectioned mesh can be calculated as the degree of matching.

[Equation 10]
If the degree of coincidence α is high (Many), the degree of the buried pipe is high (High),
(10)
[Equation 11]
If The degree of coincidence α is low (Few), and the degree of membership of the Ten buried pipe is low (Low).
(11)

Then, the degree _μgHP belonging to the buried pipe based on the knowledge of the hyperbolic shape is obtained by a known Min-Max centroid method from the membership function shown in FIG. In the input membership function of FIG. 33A, the following values are parameters for estimation processing according to the present embodiment.
(A) The degree of coincidence A1 when the membership function grade of “many” becomes 0.
(B) Boundary coincidence A2 when the grade of the membership function of “low” decreases from 1.
(C) A degree of coincidence A3 when the grade of the membership function “less” becomes zero.
(D) The flat angle A4 of the density change at the boundary when the grade of the membership function of “many” becomes 1.

That is, in the example of FIG. 33, when the degree of coincidence α is A _in , it intersects with the function lines of the “large” and “small” membership functions and extends these intersections horizontally to the output membership function. Here, the area A _H 'below the grade where the straight line extending from the intersection of “many” intersects the function curve of the membership function of “high” output, and the straight line extending from the intersection of “small” is the output “ The area centroid _μgHP in the area of the union with the area A _L ′ below the grade that intersects the function curve of the “low” membership function is obtained, and is used as the output value _μgHP of the degree of the buried pipe.

Further, the integration of knowledge of density and knowledge of hyperbolic shape will be described below. The expert of buried exploration estimates the buried pipe by weighting the above two knowledge according to the depth. Prioritize knowledge of the hyperbolic shape at shallow depths, and prioritize knowledge of density at deeper depths. Therefore, in the method according to the present embodiment, the weighting coefficient for each degree of membership is changed according to the depth as shown in FIG. Here, _{w i} is the weighting coefficient of belonging degree μ _gI, is _{w h} is the weight of belonging degree _{μ gHP.} Figure 34 is a function of the weighting coefficient for the depth, the function of the left of the trapezoidal shape is a function of the weighting coefficients w _i, a function of the right trapezoidal shape is a function of the weighting coefficient w _h. Here, in the two trapezoids, the base length (D6-D1) of the trapezoid and the length Da of the top side are set to be equal, and D1, D2, D4, Da, If the five parameter values of D6 can be determined, the function is defined. Note that the leftmost portion in FIG. 34 is a so-called “fogging” portion located in the shallow portion.

Further, by _combining the affiliation degrees μ _gI and μ _gHP obtained above and the weighting coefficients w _i and w _h as shown in the following expression, the degree of buried pipe μ _TD based on all knowledge is obtained. Then, the image data satisfying μ _TD > 0.5 is set as the image data of the buried pipe candidate.

[Equation 12]
_{μ TD = (w i × μ} gI + w h × μ gHP) / 2 (12)

  Further, the processing of each subroutine shown in FIGS. 6 to 9 for executing processing using knowledge of density, knowledge of hyperbolic shape, and the integration method thereof will be described below.

  FIG. 6 is a flowchart showing a process (step S15) of estimating a buried position candidate by searching for a minimum point of light and shade based on image data of an XZ slice of a buried position candidate, which is a subroutine of FIG. In FIG. 6, first, in step S71, a process (FIG. 7) for estimating a buried position candidate is performed by searching for the minimum point of gray based on the image data of the XZ slice of the buried position candidate after removal, and after removal in step S72. In the image data of the XZ slice of the embedded position candidate, a valley line is extracted with reference to the density, and a process of estimating the embedded position candidate by executing a hyperbolic pattern in the convex direction (FIG. 8) is executed. Further, in step S73, a process of integrating the processes for estimating the two embedded position candidates by weighting (FIG. 9) is executed, and the process returns to the original routine.

  Note that step S15 in FIG. 6 is processing for estimating the image data of the embedded position candidate based on the image data of the XZ slice of the embedded position candidate after removal, while step S25 in FIG. 5 is the process of estimating the embedded position candidate after removal. This is a process for estimating the image data of the embedded position candidate based on the image data of the XZ slice, and is executed in the same manner as the process of FIG.

FIG. 7 is a flowchart showing a process (step S71) of estimating the embedment position candidate by the gray minimum point search which is the subroutine of FIG. In FIG. 7, first, in step S81, based on the image data of the XZ slice of the embedment position candidate after removal, the angle _θaz of the density change from the minimum gray point to the periphery thereof is obtained. Then, determined by Min-Max gravity method using a membership function representing the degree mu _gI buried position candidates based on the average angle theta _az change in concentration in Figure 31 at step S82, the flow returns to the original routine.

FIG. 8 is a flowchart showing a process (step S72) of estimating the embedded position candidate by looking at the hyperbolic pattern which is the subroutine of FIG. In FIG. 8, first, in step S83, the degree of coincidence α between the hyperbola obtained by referring to the density and the theoretical hyperbola is obtained from the image data of the XZ slice of the embedded position candidate after removal. Next, the degree _μgHP of the buried position candidate based on the degree of coincidence α between the hyperbola obtained by referring to the concentration in step S84 and the theoretical hyperbola is obtained by the Min-Max centroid method using the membership function shown in FIG. Return to the original routine.

FIG. 9 is a flowchart showing a process (step S73) of integrating the process of estimating the two embedded position candidates, which is the subroutine of FIG. 6, by weighting. 9, in accordance with weighting coefficients w _h degree mu _GHP buried position candidates by weighting coefficients w _i and hyperbolic pattern degree mu _gI buried position candidate according to the minimum point exploration in the depth as shown in Figure 34 in step S85 Change and seek. Next, in step S86, the degree μ _{TD of the} buried pipe based on all knowledge is obtained by linearly combining the obtained degree and the weighting coefficient using Expression (12). Further, in step S87, image data satisfying μ _TD > 0.5 is extracted as the embedded pipe candidate image data, and the process returns to the original routine.

  10 and 11 are flowcharts showing the connection exploration process (step S5), which is a subroutine of FIG.

  In step S31 of FIG. 10, first, image data of candidate points is searched on one XY plane and stored in the processing memory 25 which is a temporary memory. Next, image data of one candidate point is selected in step S32, and the candidate point is searched for in a range sphere centered on the candidate point selected in step S33. Then, it is determined whether or not the candidate point selected in step S34 is image data of an XZ slice. If YES, the process proceeds to step S35, and if NO, the process proceeds to step S37. In step S35, it is determined whether a candidate point has been found in the image data of another XZ slice. If YES, the process proceeds to step S40, and if NO, the process proceeds to step S36. In step S36, it is determined whether a candidate point has been found in the image data of another YZ slice. If YES, the process proceeds to step S40. If NO, the process proceeds to step S42. Further, in step S37, it is determined whether or not a candidate point has been found in the image data of another YZ slice. If YES, the process proceeds to step S40, and if NO, the process proceeds to step S38. In step S38, it is determined whether a candidate point has been found in the image data of another XZ slice. If YES, the process proceeds to step S40. If NO, the process proceeds to step S42.

  In step S40, it is determined that the selected candidate point and the searched candidate point are connected, and the selected candidate point and the shortest candidate point of the searched candidate points are connected to temporarily serve as connected data. After the candidate points stored in the processing memory 25 as a memory and connected in step S41 are selected, the process returns to step S33 to perform further connection processing. Here, the concatenated data is data indicating that the image data of one candidate point and the image data of another candidate point are concatenated, and is an arrangement address of the three-dimensional image of the two candidate points. Expressed in pairs.

  It is determined whether or not there is image data of unprocessed candidate points other than those selected in step S42. If YES, the process proceeds to step S43 to select image data of another candidate point and use another embedded pipe. In order to search for a certain connected state, the process returns to step S33. On the other hand, if NO in step S42, the process proceeds to step S44 in FIG.

  In step S44 of FIG. 11, it is determined whether or not there is an unprocessed XY plane. If YES, the process proceeds to step S45, and if NO, the process proceeds to step S46. In step S45, after storing the image data of the candidate point on another XY plane in the processing memory 25 which is a temporary memory, the process returns to step S32 in FIG. 10 in order to process the image data of another XY plane. . On the other hand, an average interval is calculated for the candidate line of the embedded object in the concatenated data collected in step S46, candidate lines within the predetermined interval are determined to be the same embedded object, and a plurality of candidate lines determined to be the same embedded object. Connection data of candidate lines other than one of them is removed from the processing memory 25. Then, the concatenated data remaining in step S47 is stored in the processing memory 25 as concatenation result data, and the process returns to the original main routine.

  10 and 11, the selection of the XY plane in steps S31 and S45 is preferably performed sequentially from the exploration plane 92 in the direction of the embedding depth, and the selection of candidate points on one XY plane is the plane. It is preferable to perform processing while searching for a candidate point located at a coaxial position in the inner direction by searching for a candidate point located at a coaxial position from the outer periphery.

  FIG. 12 is a flowchart showing an embedded object extraction process (step S6) which is a subroutine of FIG.

In FIG. 12, first, one concatenation result data is selected from the concatenation result data in the processing memory 25 in step S51, and two embedded position candidates located at both ends of the concatenation result data selected in step S52 from the outer periphery. The average value of the distance and the search for a plurality of embedded position candidates of the selected connection result data is obtained. Next, in step S53, the grades G _pNear and G _pFar for the two distances from the outer periphery are calculated using the membership function of FIG. 29A, and in step S54, the membership function of FIG. _29B is used. Each grade _GpShallow , _GpMiddle , _GpDeep with respect to the average value of the _search is calculated. In step S55, the degree of affiliation grade A of the buried pipe is calculated using equation (5), and the degree of affiliation grade B of the buried pipe is calculated using equation (6). In step S56, whether grade A> grade B is satisfied. If it is YES, the process proceeds to step S57. If it is NO, the process proceeds to step S58. The connection result data selected in step S57 is extracted as an embedded pipe and stored in the image memory 23, and the process proceeds to step S59. On the other hand, the connection result data selected in step S58 is determined not to be an embedded pipe, and is not stored in the image memory 23, and the process proceeds to step S59. Further, in step S59, it is determined whether or not there is unprocessed connection result data. If YES, the process proceeds to step S60 to select another connection result data from the connection result data in the processing memory 25. The process returns to step S52. On the other hand, if NO in step S59, the buried object extraction process is terminated and the process returns to the original main routine.

  FIG. 13 is a flowchart showing a process (step S8) for adaptively controlling and setting the estimation process parameters using the genetic algorithm which is the subroutine of FIG. In FIG. 13, first, an evaluation process is executed in step S91, a selection process is executed in step S92, a crossover process is executed in step S93, a mutation process is executed in step S94, and the process returns to the original main routine. . Each processing of these steps S91 to S94 will be described in detail below.

The estimation processing parameters that need to be optimized by adaptive control are the following 14 parameters as described above.
(A) Five estimation processing parameters T1 to T5 for defining the membership function of the concentration knowledge input in FIG.
(B) Four estimation processing parameters A1 to A4 defining the membership function of the hyperbolic shape knowledge input in FIG.
(C) Five estimation processing parameters D1, D2, D4, D6, and Da that define the function of the weighting coefficient in FIG.

  In this embodiment, an individual of a gene in the genetic algorithm is expressed as an array in which 1 bit is 0,1. A different change range of each estimation processing parameter is represented by 8 bits, and for example, 10 individuals with a total of 112 bits are generated with 14 estimation processing parameters, and an initial individual population is created. The initial value of the estimation processing parameter set in step S2 is used. Here, in the present embodiment, the optimum value under the conventional single environment is assigned as each initial individual population in order to converge the optimum solution earlier. In addition, when the environment changes greatly, such as a change in the specifications of the radar device or a change in the data conversion method, it is preferable to generate an initial population at random.

When the initial individual population is generated, fitness evaluation processing (step S91) necessary for selection, crossover, and mutation operators is performed on the initial population. First, the input data used in this system is assumed as follows.
<Assumption 1> The input data includes information on all buried pipes.
<Assumption 2> There is nothing other than the buried pipe in the input data.
That is, it is assumed that all the buried pipes can be searched for the input data and there is nothing other than the buried pipe. From these assumptions, the extraction results of this system are derived as follows.
<Result 1> The extraction result includes all the buried pipes.
<Result 2> Extraction results do not exist except for buried pipes.
In this system, since the relationship between each parameter and the number of buried pipes extracted is non-linear, the fitness g is defined as follows:

Here, n is the number of extracted buried pipes, and l _i is the length of the i-th buried pipe. Since the objects extracted from <Result 2> are all buried pipes, the sum of their lengths (fitness g) is the degree of extraction of the buried pipe. When <Result 1> is satisfied, the fitness g indicates the maximum value. For example, when the buried pipe is not extracted as shown in FIG. 35 (a), the fitness g is low, and when it is accurately extracted as shown in FIG. 35 (b), the fitness g is high. In this way, the larger the total length of the buried pipe, the more accurately the buried pipe can be extracted, and the fitness g becomes higher.

In the selection process (step S92) according to the present embodiment, a known roulette selection method is used. The roulette selection method is a method in which selection probability is changed from the fitness of each individual, and selection is performed so that the roulette is rotated according to this probability. Individuals with higher fitness are more likely to be selected. When the fitness of an individual i is g _i and the number of individuals is M, selection is performed with a selection probability P _i as shown in the following equation.

FIG. 36 shows an example of selection processing by the roulette selection method used in this embodiment. Here, for example, when 10 individuals having different parameters are given as in the present embodiment, the fitness g _i is calculated using the evaluation function shown in the equation (12), Each selection probability P _i is obtained from the result. When this selection probability P _i is arranged in the roulette as shown in FIG. 36 (d), the larger the selection probability P _i is, the larger the occupation ratio is, and thus the highest selection probability is high.

Next, in the crossover process (step S93), a so-called single point crossover method is used. One-point crossover method, as shown in FIG. 37 (a) and (b), select two individuals with crossover probability p _c, determine the crossover point at random in the bit boundaries of 1 to 111, the boundary of the point Partially replace the rear gene sequence.

Furthermore, the mutation process (step S94), as shown in FIG. 38, randomly selected individuals, a bit is used to mutation mutation probability p _m was determined from the bit of 1 from 112, the bit Invert.

  FIG. 39 is a graph showing the transition of the fitness with respect to the generation number T in the embodiment of the buried object search system of FIG. In this embodiment, the number of generations (number of iterations) is changed by the loop shown in FIG. 3, and it is preferable to update the parameters until the final generation (number of iterations) T = 1000. Alternatively, the parameter update may be terminated when the rate of change decreases due to the change in the fitness of the number of generations as shown in FIG.

  As described above, according to the present embodiment, attention is paid to the hyperbolic pattern that is a characteristic of the reflected wave group from the shallow portion and the single buried object, and the feature that the amplitude drop of the reflected wave is less affected by the interference wave. The position of the buried object was estimated, and the estimation parameters were adaptively controlled using a genetic algorithm. Based on the result, a three-dimensional image of the internal structure was generated and output. As a result, it is possible to generate an image of a three-dimensional structure inside the structure with high accuracy, and as a result, it is possible to obtain information inside the structure necessary for construction. Therefore, it is possible to execute the embedded object exploration process while adaptively controlling the measurement parameters used when executing the embedded object exploration process to the optimum values, and easily and accurately search the embedded object in various exploration objects. Anyone can easily confirm the position of the buried object by visual recognition.

  In the above 2nd Embodiment, after performing a connection exploration process (step S6), the embedded object extraction process (step S7) is performed, However, this invention is not limited to this, The latter embedded object extraction It is not necessary to execute processing.

  In the second embodiment described above, in step S4, the data of the estimation processing results of the two cross sections of the XZ slice and the YZ slice are virtually allocated to be arranged in a three-dimensional space and assigned to the image memory. However, the present invention is not limited to this, and the data of the estimation processing result of at least one cross section of the XZ slice and the YZ slice is virtually combined and arranged in a three-dimensional space. Addresses may be assigned and stored in the image memory 23. In other words, the buried object exploration process may be performed based only on the cross-sectional image of one slice.

  FIGS. 40 to 42 are diagrams showing search results in the first to third embodiments of the embedded object search system of FIG. 2, respectively, and (a) of these drawings shows measurement of X-ray images according to the prior art. It is a figure which shows a result, (b) It is a figure which shows the search result image before applying the method concerning this embodiment, (c) shows the search result image after applying the method concerning this embodiment. FIG. Note that broken lines (squares) in the results illustrated in FIGS. 40B to 42B and 42C indicate portions corresponding to X-ray images.

  As is apparent from FIGS. 40 to 42, in the conventional method, the 14 estimation processing parameters of the system were fixed in advance during the three-dimensional display, so that no buried pipe in the broken line portion was detected. However, as indicated by the elliptical broken lines in FIGS. 40 to 42, the internal parameters are automatically optimized by applying the method according to the present embodiment, so that the embedding that has not been detected by the conventional method is performed. The tube could be detected.

<Modification>
In the above embodiment, when the program data of the embedded object exploration image processing of FIG. 3 is stored in the CD-ROM 45a and executed, it is loaded into the program memory 24 and executed. However, the present invention is not limited to this. , CD-R, CD-RW, DVD, MO, or other optical disk or magneto-optical disk recording medium, or a magnetic disk recording medium such as a floppy disk. These recording media are computer-readable recording media. Alternatively, the program data of the embedded object exploration image processing of FIG. 3 may be stored in advance in the program memory 24 and the image processing may be executed.

  In the above embodiment, the embedded object that continuously extends in the predetermined length direction in the exploration object, radiates the transmission wave of the electromagnetic wave from the surface of the exploration object, receives the reflected wave of the transmission wave, Although the buried object exploration processing method and apparatus for exploring based on the reflected wave data are described, the present invention is not limited to this, and the buried object searching process is similarly performed using ultrasonic waves instead of electromagnetic waves. May be.

  In the above embodiment, in the adaptive control processing of the parameter for position estimation processing using the genetic algorithm, an example is shown for each processing of evaluation, selection, crossover, mutation, and determination. It is not limited.

  As described in detail above, according to the buried object exploration processing method or apparatus according to the present invention, the buried object continuously extending in the predetermined length direction in the exploration object is converted into an electromagnetic wave or ultrasonic transmission wave. In a buried object exploration processing method or apparatus that radiates from the surface of an exploration object, receives a reflected wave of the transmitted wave, and explores based on reflected wave data, the buried object is based on the received reflected wave data. Predictive feature parameter data is extracted for exploration, and based on the extracted feature parameter data, a genetic algorithm is used to estimate the position estimation processing parameter so that the degree of fitness of the position estimation processing parameter is high The position of the buried object is estimated using the knowledge that the buried object continuously extends while adaptively controlling the parameters for use, and the position of the buried object is determined based on the estimated position of the buried object. And it generates and outputs an image in the form of a 3-dimensional image. Therefore, the embedded object exploration process can be executed while adaptively controlling the measurement parameters used when executing the embedded object exploration process to the optimum values, and the embedded object in various exploration objects can be easily compared with the prior art. Thus, it is possible to search with high accuracy, and anyone can easily confirm the position of the buried object by visual recognition.

  Further, according to the buried object exploration processing method or apparatus according to the present invention, the buried object continuously extending in the predetermined length direction in the exploration object is transmitted from the surface of the exploration object by electromagnetic waves or ultrasonic waves. In a buried object exploration processing method or apparatus for radiating, receiving a reflected wave of the transmitted wave, and exploring based on reflected wave data, orthogonal to the surface of the exploration object based on the received reflected wave data Generating image data having density data of at least one cross section, and using the knowledge that the embedded object extends continuously, removing image data having no embedded object from the generated image data Candidate point image data of at least one cross section indicating a candidate point of the embedded object is generated, and the generated at least one cross section candidate point image data is virtually arranged in a three-dimensional space. The position estimation parameters are adaptively controlled using genetic algorithms based on the candidate image data stored in the image memory and using the genetic algorithm. However, the position of the buried object is estimated by connecting candidate points using knowledge that the buried object continuously extends, and based on the estimated position of the buried object, An image is generated and output in the form of a three-dimensional image. Therefore, the embedded object exploration process can be executed while adaptively controlling the measurement parameters used when executing the embedded object exploration process to the optimum values, and the embedded object in various exploration objects can be easily compared with the prior art. Thus, it is possible to search with high accuracy, and anyone can easily confirm the position of the buried object by visual recognition.

It is a flowchart which shows the measurement process provided with the adaptive control process of the measurement parameter which is the 1st Embodiment which concerns on this invention. It is a block diagram which shows the structure of the buried object search system which is 2nd Embodiment which concerns on this invention. It is a flowchart which shows the buried object search image process (main routine) performed by CPU20 of the image processing apparatus 10 of FIG. It is a flowchart which shows the burying position candidate estimation process (step S2) based on the reflected wave signal data at the time of the scanning of the X direction which is a subroutine of FIG. It is a flowchart which shows the embedment position candidate estimation process (step S3) based on the reflected wave signal data at the time of the scanning of the Y direction which is a subroutine of FIG. FIG. 5 is a flowchart showing processing (step S15) for estimating a buried position candidate by searching for a minimum point of light and shade based on image data of an XZ slice of a buried position candidate, which is a subroutine of FIG. It is a flowchart which shows the process (step S71) which estimates an embedment position candidate by the local minimum point search of the light and shade which is the subroutine of FIG. It is a flowchart which shows the process (step S72) which estimates an embedment position candidate by seeing the hyperbolic pattern which is a subroutine of FIG. It is a flowchart which shows the process (step S73) which integrates the process which estimates the said 2 embedment position candidates which is a subroutine of FIG. 6 by weighting. It is a flowchart which shows the 1st part of the connection search process (step S5) which is a subroutine of FIG. It is a flowchart which shows the 2nd part of the connection search process (step S5) which is a subroutine of FIG. It is a flowchart which shows the buried object extraction process (step S6) which is a subroutine of FIG. It is a flowchart which shows the process (step S8) which adaptively controls and sets the parameter for estimation processes using the genetic algorithm which is a subroutine of FIG. It is a spectrum figure which shows the frequency spectrum of the transmission wave signal used in the buried object search system of FIG. It is a signal waveform diagram which shows the waveform of the transmission pulse of the transmission wave signal of FIG. It is a top view which shows the scanning method of the search used with the buried object search system of FIG. It is a front view which shows the measurement principle of the exploration apparatus provided with the transmission / reception apparatus for buried object search used with the buried object search system of FIG. It is the figure and the photograph of an image which show the hyperbolic pattern of the reflected wave group measured by the buried object search system of FIG. (A) is a waveform diagram of a received waveform measured by the buried object search system of FIG. 2, (b) is a photograph of the X direction search result image, and (c) is the Y direction search result image. It is a photograph of. (A) is a photograph of the exploration result image measured by the buried object exploration system of FIG. 2, and (b) is a waveform diagram showing the waveform on the La line in (a). (A) is a photograph of the exploration result image measured by the buried object exploration system of FIG. 2, (b) is a waveform diagram of the waveform on the Lb line of (a), and (c) is the mutual interference portion. It is an enlarged view. It is a graph which shows the density | concentration distribution measured by the buried object search system of FIG. It is a figure which shows the extraction result of the candidate area | region measured by the buried object search system of FIG. 2, Comprising: (a) is the photograph of the original image, (b) is a figure which shows the image of the processing result. (A) And (b) is a wave form diagram which shows the signal waveform each measured by the buried object search system of FIG. 1, (c) is a figure which shows the image before a process, (d) is a figure after a process. It is a figure which shows an image. (A) is a figure which shows the image of the labeling result measured by the buried object search system of FIG. 2, (b) is a figure which shows the image of the buried position estimation result. FIG. 20 is a three-dimensional perspective view showing an outline of a region N4 of FIG. 19 measured by the buried object search system of FIG. It is the figure of an image and the three-dimensional perspective view which show the example which cannot estimate the embedment position by the minimum point search measured by the embedment search system of FIG. It is a perspective view which shows the connection process of the estimated position performed by the buried object search system of FIG. It is a membership function for buried pipe extraction used in the buried object exploration system of FIG. 2, wherein (a) is a membership function graph related to the distance from the outer periphery, and (b) is a membership function graph related to the depth. It is. It is a figure which shows the characteristic of the density | concentration used in the buried object search system of FIG. 2, Comprising: (a) is the grayscale image of the search result which concerns on the reflected wave group of electromagnetic waves which showed the propagation time with respect to the investigation distance in the case of a buried pipe, and And (b) is a graph showing the density with respect to the survey distance in the grayscale image, and (b) shows the grayscale image of the search result relating to the reflected wave group of the electromagnetic wave, showing the propagation time with respect to the survey distance in the interference region. It is a figure which shows the graph which shows the density | concentration with respect to the investigation distance in the gray image. FIG. 2 is a membership function of knowledge of concentration used in the embedded object search system of FIG. 2, wherein (a) is a graph of an input membership function, and (b) is a graph of an output membership function. It is a figure which shows the characteristic of the hyperbola shape used in the buried object search system of FIG. 2, Comprising: (a) is a figure which shows the grayscale image of the search result which concerns on the reflected wave group of electromagnetic waves which showed the propagation time with respect to investigation distance. (B) is a figure which shows the hyperbola shape in the gray image of the search result, (c) is a figure which shows the theoretical hyperbola. 2 is a hyperbolic shape knowledge membership function used in the embedded object search system of FIG. 2, wherein (a) is a graph of an input membership function, and (b) is a graph of an output membership function. Weighting factors w _i, for use in the process of step S85 in FIG. 9 to the depth used in buried objects exploration system of Figure 2 is a graph showing a function of w _h. It is a figure which shows the evaluation method using the fitness in the evaluation process (step S91) of FIG. 13, Comprising: (a) is a figure which shows the search result image of an embedded pipe when a fitness is low, (b) FIG. 10 is a diagram showing a search result image of an embedded pipe when the fitness is high. It is a figure which shows the selection process using the roulette selection method in the selection process (step S92) of FIG. 13, Comprising: (a) is a figure which shows an example of each individual data in a selection process, (b) It is a figure which shows the fitness g with respect to an individual, (c) is a figure which shows the selection probability P with respect to said each individual, (d) is a figure which shows the pie chart of the said roulette selection probability. It is a figure which shows an example of the one point crossover process in the crossover process (step S93) of FIG. 13, (a) is a figure which shows the data of two selected solids before the said crossover process, (b) is the said crossover. It is a figure which shows the data of two selected solids after a process. It is a figure which shows an example of the process in the mutation process (step S93) of FIG. It is a graph which shows transition of the adaptability with respect to the generation number T in the Example of the embedded object search system of FIG. It is a figure which shows the search result in 1st Example of the buried object search system of FIG. 2, Comprising: (a) is a figure which shows the measurement result of the X-ray image by a prior art, (b) It concerns on this embodiment It is a figure which shows the search result image before applying a method, (c) is a figure which shows the search result image after applying the method which concerns on this embodiment. It is a figure which shows the search result in 2nd Example of the buried object search system of FIG. 2, Comprising: (a) is a figure which shows the measurement result of the X-ray image by a prior art, (b) It concerns on this embodiment It is a figure which shows the search result image before applying a method, (c) is a figure which shows the search result image after applying the method which concerns on this embodiment. It is a figure which shows the search result in 3rd Example of the buried object search system of FIG. 2, Comprising: (a) is a figure which shows the measurement result of the X-ray image by a prior art, (b) It concerns on this embodiment It is a figure which shows the search result image before applying a method, (c) is a figure which shows the search result image after applying the method which concerns on this embodiment.

Explanation of symbols

1 ... Electromagnetic wave transmitter / receiver for exploring buried objects,
1R ... wheel,
1RR ... car body,
2 ... Transmitting antenna,
3 ... receiving antenna,
4 ... Transmitter,
5 ... Receiver,
6 ... Controller,
7 ... Communication interface,
10 Image processing device,
20 ... CPU,
21 ... ROM,
22 ... RAM,
23. Image memory,
24 ... Program memory,
25 ... Processing memory,
26: Reception memory,
30 ... Bus
31 ... Keyboard interface,
32 ... Mouse interface,
33 ... Display interface,
34 ... Printer interface,
35 ... Drive device interface,
41 ... Keyboard,
42 ... mouse,
43 ... CRT display
44 ... Printer,
45 ... CD-ROM drive device,
45a ... CD-ROM,
51. Communication cable,
61: Communication interface.
90 ... buried object,
91 ... exploration object,
92: Exploration surface.

Claims (10)

Reflected wave data is obtained by radiating an electromagnetic wave or ultrasonic transmission wave from the surface of the exploration object, receiving the reflected wave of the transmission wave, and the embedded object continuously extending in the predetermined length direction in the exploration object. In the buried object exploration processing method for exploring based on
Based on the received reflected wave data, image data having density data of at least one cross section orthogonal to the surface of the exploration object is generated, and using the premise that the embedded object extends continuously. Generating candidate point image data of at least one cross section indicating candidate points of the embedded object by removing image data in which no embedded object exists from the generated image data;
Storing the generated candidate point image data of at least one cross section in an image memory so as to be virtually arranged in a three-dimensional space;
Based on the candidate point image data stored in the image memory, using a genetic algorithm, the embedded object is controlled while adaptively controlling the position estimation processing parameters so that the degree of fitness of the position estimation processing parameters is high. Estimating the position of the buried object by connecting candidate points using the premise that they extend continuously;
Based on the position of the estimated buried object, it viewed including the step of generating and outputting an image of the buried object in a three-dimensional image format,
The step of estimating the position of the buried object is as follows:
The knowledge of the density in the candidate point image data stored in the image memory is expressed, and a predetermined first membership function is used to indicate the degree of the buried position candidate based on the average angle of density change. Calculating the degree,
Represents knowledge of hyperbolic shape in candidate point image data stored in the image memory, and indicates the degree of the embedded position candidate based on the degree of coincidence between the hyperbola obtained by referring to the density in the candidate point image data and the theoretical hyperbola. Calculating a degree of the second embedded object using a predetermined second membership function shown;
The degree of the third buried object is calculated by integrating the calculated degree of the first buried object and the calculated degree of the second buried object by linear combination using a function of a predetermined weighting coefficient. Steps,
And a step of estimating the position of the embedded object by determining whether or not the image data is an embedded object candidate based on the degree of the third embedded object . In the buried object exploration processing method according to claim 1 ,
After the step of estimating the position of the buried object,
(A) the first rule that both ends of the buried object are close to the end of the exploration object, and the depth of the buried object is not between the cover thicknesses of the exploration object layer, the probability of being an embedded object is high;
(B) a second rule that the probability that the object is a buried object is low when both ends of the buried object are far from the end of the exploration object and the depth of the buried object is between the cover thicknesses of the exploration object layer. Using fuzzy rules,
Based on the estimated position of the embedded object, it is determined whether or not the estimated position of the embedded object is an embedded object, and when the estimated position of the embedded object is determined to be not an embedded object, The embedded object exploration processing method further comprising the step of removing from the data. 3. The embedded object exploration processing method according to claim 1, wherein the position estimation processing parameters are a parameter for defining the first membership function and a parameter for defining the second membership function. And a parameter for defining the function of the weighting coefficient. In buried objects exploration method according to one of claims 1 to 3, the goodness of fit, the length of each buried pipe which is extracted by estimating the candidate point image data stored in the image memory A buried exploration processing method characterized by the summation. Reflected wave data is obtained by radiating an electromagnetic wave or ultrasonic transmission wave from the surface of the exploration object, receiving the reflected wave of the transmission wave, and the embedded object continuously extending in the predetermined length direction in the exploration object. In the buried object exploration processing equipment to explore based on
Based on the received reflected wave data, image data having density data of at least one cross section orthogonal to the surface of the exploration object is generated, and using the premise that the embedded object extends continuously. Generating means for generating candidate point image data of at least one cross section indicating candidate points of the embedded object by removing image data in which no embedded object exists from the generated image data;
Storing means for storing the generated candidate point image data of at least one cross section in an image memory so as to be virtually arranged in a three-dimensional space;
Based on the candidate point image data stored in the image memory, using a genetic algorithm, the embedded object is controlled while adaptively controlling the position estimation processing parameters so that the degree of fitness of the position estimation processing parameters is high. Position estimation means for estimating the position of the buried object by connecting candidate points using the premise that it extends continuously;
Based on the position of the estimated buried object, see containing and output means for generating and outputting an image of the buried object in the form of a 3-dimensional image,
The position estimating means includes
The knowledge of the density in the candidate point image data stored in the image memory is expressed, and a predetermined first membership function is used to indicate the degree of the buried position candidate based on the average angle of density change. A first calculating means for calculating the degree;
Represents knowledge of hyperbolic shape in candidate point image data stored in the image memory, and indicates the degree of the embedded position candidate based on the degree of coincidence between the hyperbola obtained by referring to the density in the candidate point image data and the theoretical hyperbola. Second calculating means for calculating a degree of the second embedded object using a predetermined second membership function shown;
The degree of the third buried object is calculated by integrating the calculated degree of the first buried object and the calculated degree of the second buried object by linear combination using a function of a predetermined weighting coefficient. A third calculating means;
An embedded system comprising: embedded position estimation means for estimating the position of the embedded object by determining whether the image data of the embedded object candidate is based on the degree of the third embedded object. Object exploration processing equipment. In the buried object exploration processing device according to claim 5 ,
After the processing of the position estimation means,
(A) the first rule that both ends of the buried object are close to the end of the exploration object, and the depth of the buried object is not between the cover thicknesses of the exploration object layer, the probability of being an embedded object is high;
(B) a second rule that the probability that the object is a buried object is low when both ends of the buried object are far from the end of the exploration object and the depth of the buried object is between the cover thicknesses of the exploration object layer. Using fuzzy rules,
Based on the estimated position of the embedded object, it is determined whether or not the estimated position of the embedded object is an embedded object, and when the estimated position of the embedded object is determined to be not an embedded object, A buried object exploration processing apparatus, further comprising a removal means for removing data. 7. The embedded object exploration processing apparatus according to claim 5, wherein the position estimation processing parameter includes a parameter for defining the first membership function and a parameter for defining the second membership function. And a parameter for defining the function of the weighting coefficient. In buried object exploration processing apparatus according to one of claims 5 to 7, the goodness of fit, the length of each buried pipe which is extracted by estimating the candidate point image data stored in the image memory A buried object exploration processing device characterized by a sum. Buried object exploration program, which comprises the steps of buried objects exploration method according to one of claims 1 to 4. A recording medium in which the buried object search processing program according to claim 9 is recorded.

Images