KR101895835B1 - Ground penetrating radar survey system - Google Patents

Abstract

The present invention relates to a ground penetrating radar survey system including an image acquisition unit including an information processing unit processing information collected by an information collection unit in an information transmission and reception unit and an information display unit displaying the same and provided with a lens adapter and a digital camera installed in the front, side, and rear end portions of a pupil survey car to simultaneously acquire images of the front of the ground surface, a lateral ground object, and the ground surface related to conduit and pupil-related positional information; a ground penetrating radar disposed in the lower portion of the pupil survey car for conduit and pupil information acquisition; an encoder installed on one front side of the ground penetrating radar and measuring the traveling distance and the traveling speed of the pupil survey car in a case where the pupil survey car travels on the upper surface of the conduit; and a module-based intelligent image analysis device incorporating an inspection and analysis program synthesizing, editing, and analyzing an image by using data collected through a vibration frequency characteristic extraction module collecting and storing electromagnetic wave data obtained from the ground penetrating radar and image data obtained from the image acquisition unit.

Description

{GROUND PENETRATING RADAR SURVEY SYSTEM}

More particularly, the present invention relates to an information transmission unit for processing information collected by an information collecting unit in an information transmitting and receiving unit and an information display unit for displaying the same, An image acquiring unit installed at an end of the image capturing unit and equipped with a lens adapter and a digital camera for simultaneously acquiring front and side topographical images and a ground image on the ground related information of the pipeline and the pupil; An indicator transmission radar installed at a lower portion of the pupil exploration vehicle to acquire pipeline and pupil information; An encoder installed at a front side of the surface transmission radar and measuring a travel distance and a traveling speed of the pupil-searching vehicle when the pupil-surveying vehicle runs on a channel upper surface; An analysis program for synthesizing, editing, and analyzing images using data collected through an oscillation frequency feature extraction module for collecting and storing the image data acquired by the image acquisition unit and the electromagnetic wave data acquired from the surface transmission radar And an embedded module based intelligent image analysis apparatus.

In recent years, road accidents due to aging of the city have been frequent and citizens 'anxiety has been increasing due to the loss of citizens' lives and property.

In August 2014, five road depressions occurred near the Seokchon lake in Seoul, causing the urban road depression to become a social issue. In February 2015, a middle school student who was walking along the road near Yongsan Station construction site was injured in a road accident The government, local governments, academia, and industry are preparing various countermeasures, but puppets are constantly emerging.

The pupil density in Seoul is concentrated in the gneissic area and in the alluvial area of the Han river. The pupil occurrence rate is highest in Seoul and the number of pupil incidence is continuously increasing since 2011.

 The existing single channel GPR probe consists of one transmitter and one receiver, and it is possible to acquire only one profile in one measurement.

However, it was insufficient to find the hidden pupil at the bottom of the road, and did not provide a multifaceted and effective method for the persistent pupil.

Especially, it is very difficult to observe the pupil precisely even if the equipment with high performance is used.

Korean Patent Publication No. 2016-0038595 Korean Patent No. 1662050 Korea Patent Publication No. 2011-0052885 Korean Patent Publication No. 2016-0095273

SUMMARY OF THE INVENTION It is an object of the present invention to provide a ground penetrating radar detection system capable of detecting a pupil in advance and performing safety measures in order to prevent road depression caused by a hypocentral pupil below the road surface in Seoul The purpose is to provide.

In addition, since the present invention is installed in a vehicle and has 25 channels, it is possible to acquire 25 longitudinal profiles, planes, and cross sections in a single search, save time and cost compared to the existing single channel, The objective is to provide an exploration system.

In order to solve the above-described problems, the present invention provides a lens adapter and a digital camera installed at the front, side, and rear ends of a pupil-seeking vehicle for simultaneously acquiring front and side surface features and ground images on the ground An image acquisition unit; An indicator transmission radar installed at a lower portion of the pupil exploration vehicle to acquire pipeline and pupil information; An encoder installed at a front side of the surface transmission radar and measuring a travel distance and a traveling speed of the pupil-searching vehicle when the pupil-surveying vehicle runs on a channel upper surface; And an oscillation frequency acquisition module installed at a lower portion of the pupil-seeking vehicle to acquire a vibration frequency.

An information processing unit incorporating a survey analysis program for collecting and storing the image data acquired by the image acquisition unit and the electromagnetic wave data acquired from the indicator transmission radar and synthesizing, editing, and analyzing the images using the collected data; And

And an information display unit for displaying information processed by the information processing unit,

The information processing unit extracts Mel-Frequency Cepstrum Coefficient (MFCC), Time domain features, and Frequency domain features from the vibration vibration information acquired by the vibration frequency acquisition module, A vibration frequency feature extraction module for extracting a vibration frequency characteristic of the vibration frequency;

A pupil recognition module for detecting and classifying pupils by inputting vibration frequencies acquired in real time from the vibration frequency feature extraction module through a previously learned DNN model as a DNN (Deep Neural Network) which is one of deep learning algorithms.

In order to avoid the over-fitting of the DNN model for the pupil information detected and classified by the pupil recognition module and to avoid the mutual adaptation between the nodes of the neural network model, A pupil detection and classification module for re-detecting and re-classifying the pupils by learning after excluding the pupils;

Wherein the vibration frequency characteristic extracted from the vibration frequency feature extraction module is optimized by using a correlation-based feature selection (CFS) or a convolutional neural network (CNN) And an optimization module.

A monitoring module installed at a lower side of the manhole cover to detect and communicate the pupil state; A vibration sensor installed in the monitoring module to generate vibration information; An antenna unit for communication; A manhole cover and a control unit for collecting information on the manhole and its periphery through the vibration sensor unit and transmitting manhole information including vibration information and information on the periphery of the manhole to an information processing unit through an antenna unit, The connected information processing unit receives the information of the manhole and its periphery with a radio signal from the controller, and detects, classifies, and controls the pupil.

The present invention as described above has the effect of protecting the safety and life of the citizens by performing pupil detection before the occurrence of the road depression due to the pupil and taking preliminary measures against damage to the pupil.

It is effective to improve the puppy detection technology in the early stage by expanding the industry of new industries.

The vehicle-type multi-channel antenna according to the present invention is composed of 25 channels mounted on a vehicle, so that it is possible to easily obtain 25 cross-sections for each of the longitudinal cross-sectional distances of 2.4 m, which is larger than that of the conventional single-

Further, the present invention has a faster searching speed than the conventional single channel, and has a time and cost reduction effect.

In addition, the present invention can identify the pupils before the road depression with accurate pupil recognition and protect the safety and property of citizens by performing emergency repair work in case of dangerous pupil.

In addition, the present invention is capable of taking precautions at the time of road pavement and structure construction of a city by grasping pupil generation mechanism by accumulating and analyzing data such as pupil position information.

Further, the present invention has the effect of establishing a dangerous pupil management method suitable for characteristics of a road pavement layer by accumulating pupil-related data.

BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a diagram showing the concept of an indicator transmission radar survey system according to the prior art.
2 is a view showing a concept of an indicator transmission radar survey system according to an embodiment of the present invention.
3 is a configuration diagram according to an embodiment of the present invention.
4 is a diagram illustrating a pupil waveform analysis technique according to an embodiment of the present invention.
5 is a diagram illustrating application of an optimal pupil analysis program according to an embodiment of the present invention.
FIG. 6 is a diagram showing utilization of a pupil waveform DB data of the surface transmission radar survey system according to an embodiment of the present invention.
FIG. 7 is a view showing an apparatus for enlarging the width of a surveying radar surveying system according to an embodiment of the present invention.
FIG. 8 is a view showing that a speed of 80 km / h is possible by improving the road surface image acquisition speed according to an embodiment of the present invention.
FIG. 9 is a view showing the configuration of FIG. 7 in detail according to an embodiment of the present invention.

For a better understanding of the present invention, a preferred embodiment of the present invention will be described with reference to the accompanying drawings.

The embodiments of the present invention may be modified into various forms, and the scope of the present invention should not be construed as being limited to the embodiments described in detail below. The present embodiments are provided to enable those skilled in the art to more fully understand the present invention. Therefore, the shapes and the like of the elements in the drawings can be exaggeratedly expressed to emphasize a clearer description.

It should be noted that in the drawings, the same members are denoted by the same reference numerals.

Further, detailed descriptions of well-known functions and configurations that may be unnecessarily obscured by the gist of the present invention are omitted.

Since the present invention is composed of 25 channels mounted on a vehicle, it is possible to easily acquire 25 cross-sections for each of the longitudinal cross-sectional distances of 2.4 m, which is wider than the conventional single-channel antenna, thereby improving the analytical technique for accurately detecting all the pupil of the probe route.

Therefore, as one embodiment of the present invention, a high hit rate can be expected in the pupil examination by reducing the pupil waveform error rate.

3, the information processing unit 13 displays information collected by the information collecting unit 12 in the information transmitting and receiving unit 11, The pupil detection information can be collected and the damage to the pupil can be taken in advance.

The information processing unit 13 may use a method such as a Correlation-based Feature Selection (CFS) or a Deep Learning method, such as a Convolutional Neural Network (CNN) , And uses one of the machine learning and deep learning algorithms for pupil detection and classification.

In addition, a module-based intelligent image analyzing apparatus for analyzing a region of interest around a pupil in a module-based intelligent image analyzing apparatus that receives and analyzes an image from a wired / wireless image tracking apparatus installed in the information processing unit 13, It is composed of input image edge extraction function, horizontal edge grouping function, interest area fence automatic setting function, and Min-Max background modeling and GMM fusion by the Soebel operator for background detection and motion detection The area noise control function, the moving object extracting function, and the HOG modeling method, and displayed on the information display unit 14. [

That is, the pupil recognition module installed in the information processing unit 13 performs a background separation algorithm function for detecting a moving object by merging Min-Max background modeling and GMM so that pupil can be detected except the background in the pupil from the input image, It is a module that includes a region extraction function, a pupil region noise control function, and a pupil position extraction function, so that the position and size of the pupil of the correct shape can be known.

Further, the pupil recognition module can detect a pupil using methods such as a correlation selection method such as CFS (Correlation-based Feature Selection) or a method such as CNN (Convolutional Neural Network), which is one of deep learning methods And classification.

In addition, as shown in FIG. 5, the present invention can employ a GPR IS program capable of simultaneously checking a GPR plane, an end point, a traverse, an acquired waveform, a road surface, and a surrounding image.

Therefore, according to the present invention, it is possible to enlarge the scanning width to 2.4 m, thereby eliminating the pupil-unidentifiable section.

As shown in FIG. 6, it is possible to classify the waveform characteristics of the pupil and the position into DBs, thereby improving the pupil discrimination analysis capability.

6, the present invention includes a sound acquisition module, a vibration frequency feature extraction module, a vibration frequency feature optimization module, a fault detection and classification module Module).

The vibration frequency acquisition module acquires a vibration frequency generated in the underground probe from the radar sensor.

For this purpose, the vibration frequency information at the time of underground exploration is effectively acquired from the sensor installed near the pupil. At this time, it is possible to accurately acquire the actual pupil state information without external constraints unlike the method of directly monitoring by the person.

Therefore, since the pupil can be detected and classified by the sensor alone, the initial cost is low, and the calculation amount can be reduced by reducing the dimension of the vibration frequency characteristic vector.

In addition, by the vibration frequency feature extraction described below, it is possible to grasp the pupil of the part where the analysis by the vibration frequency information can not be observed apparently, rather than the analysis by the image processing.

It is also possible to use a publicly available benchmark vibration frequency dataset with qualitative analysis to enhance the performance of the unsteady vibration frequency detection and identification system even in the harsh environment of the external ground.

The vibration frequency feature extraction module extracts various vibration frequency characteristics (MFCC: Mel-Frequency Cepstrum Coefficient, time domain features, and frequency domain features, etc.) from the probe vibration information acquired by the vibration frequency acquisition module.

For example, the vibration frequency feature extraction module extracts various vibration frequency characteristics from the pupil vibration frequency inputted from the vibration frequency acquisition module.

The extracted frequency-frequency characteristics extract the characteristics of the Mel-Frequency Cepstral Coefficients (MFCC), which is a method of expressing the power spectrum of a short-term signal mainly used.

Here, MFCC is a cepstrum coefficient extraction method that reflects the vibration characteristics along the mel scale.

After performing the FFT, the vibration frequency signal is represented by a spectrum, and the filter bank and log are calculated in units of melts.

After performing the logarithm calculation with the Melt filter bank, DCT (Discrete Cosine Transform) is performed to extract MFCC coefficients composed of a plurality of frequency characteristics and one frame log energy characteristic.

As an embodiment of the present invention, in order to detect and identify an abnormal situation using the vibration frequency information, an abnormal vibration frequency that can be generated at each place where the vibration sensor is installed is defined in advance, and is defined as a frequency frequency coefficient Is expressed as a characteristic value that is converted numerically. Then, each unsteady vibration frequency is learned by using a machine learning algorithm such as GMM, HMM (Hidden Markov Model), SVM, etc., and converted into a feature value used for learning by inputting the vibration frequency signal input every second, You can also detect it.

The characteristics of the vibration frequency to be used in the detection and classification described below may utilize not only MFCC but also Mel-Log Spectrum, Time Domain characteristic, and Frequency Domain characteristic. In addition, the vibration frequency to be used for detection and classification includes a preprocessing function for noise and the like.

The vibration frequency characteristic optimization module selects an important vibration frequency characteristic vector capable of detecting the pupil by applying the vibration frequency characteristics extracted from the vibration frequency characteristic extraction module to a characteristic subset selection algorithm such as CFS (Correlation Feature Selection).

Such optimization of the vibration frequency characteristic can reduce the calculation amount by reducing the dimension of the vibration frequency characteristic vector in order to reduce the calculation time of the pupil detection and classification in the pupil detection and classification module, and at the same time, By selecting important features that affect it, you can ensure stable performance.

The pupil detection and classification module uses one of a machine learning and a deep learning algorithm for pupil detection and classification.

For example, the SVM (Support Vector Machine) based on the statistical learning theory shows excellent performance in the pupil pattern recognition field by converting a given problem into a convex quadratic problem which is always guaranteed as a global optimal solution.

Here, the abnormal situation detection problem can be regarded as a binary class classification problem that distinguishes between normal and abnormal situations (abnormality). In the present invention, SVM (Support Vector Machine), which is an effective machine learning method for classifying a binary class, is used as an algorithm for detecting an abnormal situation.

Also, for example, in the case of the DNN (Deep Neural Network), which is one of the deep learning algorithms, the pupil recognition module can detect and classify the pupil by inputting the vibration frequencies acquired in real time through the DNN model previously learned .

In the DNN learning, a feature vector obtained by reducing the dimension obtained by the vibration frequency feature optimizing module is input, and can be classified into various states according to various states of the pupil.

The DNN model used in the pupil detection and classification module is a neural network model having a structure having at least one hidden layer between the input layer and the output layer unlike the conventional artificial neural network.

The deep neural network with one or more hidden layers increases learning capacity including a large number of weights, so that a larger number of learning data can be used to learn more information.

In addition, in the pupil detection and classification module, in order to avoid over-fitting of the DNN model and to avoid mutual adaptation between nodes of the neural network model, a dropout technique is used to detect any normal nodes ), And then proceeds to study

The pupil detection and classification module also detects and classifies the pupil in real time from feature information of the vibration frequency acquired for the pupil using various other machine learning algorithms.

Or acquires only the pupil portion using the vibration frequency acquisition process algorithm with the vibration frequency signal input from the vibration sensor, calculates a value according to the amplitude of the region in a specific frequency region of the normal or abnormal vibration frequency, If the specific region of the frequency of the vibration is larger than the predetermined amplitude threshold value, the threshold comparison method may be used.

Therefore, through the various machine learning algorithms, not only the pupil detection but also the classification of the pupil can be anticipated.

Further, as another embodiment of the present invention, pupil is detected in real time from the feature information of the vibration frequency obtained from the pupil using various machine learning algorithms, and a report on the situation on the spot It is possible to input information about "coincidence" or "mismatch" in the check item, so that the pupil is detected in real time from the feature information of the vibration frequency and if the item is not matched with the classified pupil, the manager writes the pupil state manually Or stored in the database so that they can be searched again.

As shown in FIG. 8, the speed of road surface image acquisition can be improved to perform the 80 km / h exploration.

For example, if a vehicle GPR is inserted first and analyzed as a pupil (eg, a sinkhole), a hand-held GPR is used to confirm the presence of a sink hole in the suspected area, and a hole is drilled to confirm the sink hole with the endoscope . Vehicle-mounted GPR equipment can survey roads up to 2.4 m wide at a speed of 80 km / h. Endoscopy equipment is a state-of-the-art equipment that 360 ° rotates underground and records sections.

As shown in Fig. 9, it is possible to eliminate the curved portion positioning and the measurement distance measurement error through the vehicle-integrated type, odometer.

The onboard GPR of the present invention can include an alarm device in an external display device. A basic alarm is set. When a condition is changed, an alarm according to a condition is transmitted to an information processing unit. In addition, Generates a special alarm that contains information about a special situation.

Here, the alarm according to the special situation can be set such that, for example, if the manager who is the first alarm target is absent or is not contacted within a predetermined time, the second and third administrators can be contacted in turn, So that you can control not only the alarm but also the method and process.

In addition, alarm contents can be divided into exploration> observation> safety grade, △ emergency recovery △ first recovery △ general recovery △ observation object.

In addition, if 'emergency repair application' is installed in the terminal of the administrator, it is possible to receive the report of road drowning (including road breakage) to the 'port hall real-time reporting system' → simultaneous delivery of public officials, supervisor, .

For example, by adjusting the schedule schedule of the managers, the maximum number of alarms is adjusted to the manager terminal so that the number of alarms is reduced in the case of leaving the specific manager, and the number of alarms is adjusted to be less than a predetermined value,

 The information processing unit is configured to determine whether or not the set of the execution command set 1 and the set of the execution command set 2 match the set of the execution command set 2, The operation is reflected by reflecting the failure occurrence countermeasure information of the execution instruction set 1.

9, the pupil probe of the present invention may include various cameras, a main antenna, a traction wire, and the like. In addition, the pupil probe may be installed at the front, side, and rear ends of the pupil- An image acquiring unit including a lens adapter and a digital camera for simultaneously acquiring front, side, and topographic images on the information related paper; An indicator transmission radar installed at a lower portion of the pupil exploration vehicle to acquire pipeline and pupil information; An encoder installed at a front side of the surface transmission radar and measuring a travel distance and a traveling speed of the pupil-searching vehicle when the pupil-surveying vehicle runs on a channel upper surface; An analysis program for synthesizing, editing, and analyzing images using data collected through an oscillation frequency feature extraction module for collecting and storing the image data acquired by the image acquisition unit and the electromagnetic wave data acquired from the surface transmission radar Embedded module based intelligent image analysis device; A transmitting antenna capable of radiating an electromagnetic wave toward an indicator in a state of being spaced apart from an indicator; And a reception antenna that is capable of receiving the electromagnetic waves reflected from the target after being radiated from the transmission antenna in a state of being separated from the surface of the target.

As one embodiment of the present invention, there is provided a portable information terminal including a lens adapter and a digital camera provided at a front, a side, and a rear end of a pupil-seeking vehicle for simultaneously acquiring front and side topographic images and a ground image on a ground- An acquisition unit; An indicator transmission radar installed at a lower portion of the pupil exploration vehicle to acquire pipeline and pupil information; An encoder installed at a front side of the surface transmission radar and measuring a travel distance and a traveling speed of the pupil-searching vehicle when the pupil-surveying vehicle runs on a channel upper surface; And an oscillation frequency acquisition module installed at a lower portion of the pupil-seeking vehicle to acquire a vibration frequency.

An information processing unit incorporating a survey analysis program for collecting and storing the image data acquired by the image acquisition unit and the electromagnetic wave data acquired from the indicator transmission radar and synthesizing, editing, and analyzing the images using the collected data; And

And an information display unit for displaying information processed by the information processing unit,

The information processing unit extracts Mel-Frequency Cepstrum Coefficient (MFCC), Time domain features, and Frequency domain features from the vibration vibration information acquired by the vibration frequency obtaining module, A vibration frequency characteristic extraction module for extracting a vibration frequency characteristic;

A pupil recognition module for detecting and classifying pupils by inputting vibration frequencies acquired in real time from the vibration frequency feature extraction module through a previously learned DNN model as a DNN (Deep Neural Network) which is one of deep learning algorithms.

In order to avoid the over-fitting of the DNN model for the pupil information detected and classified by the pupil recognition module and to avoid the mutual adaptation between the nodes of the neural network model, A pupil detection and classification module for re-detecting and re-classifying the pupils by learning after excluding the pupils;

Wherein the vibration frequency characteristic extracted from the vibration frequency feature extraction module is optimized by using a correlation-based feature selection (CFS) or a convolutional neural network (CNN) And an optimization module.

In an embodiment of the present invention, it is preferable that the monitoring modules are installed in every manhole cover provided at regular intervals on the road.

A monitoring module installed on a lower side of a manhole cover for sensing the pupil state and detecting and communicating the state of the inside of the manhole and the periphery thereof; A vibration sensor unit installed in the monitoring module; An antenna unit for communication; A control unit for collecting information on the manhole cover and information on manholes and surrounding areas through the vibration sensor unit and transmitting information on the manhole including the vibration information and its peripheral part to the information processing unit through the antenna unit; The information processing unit is connected to the manhole cover and the network. The information processing unit receives the information of the manhole and the periphery of the manhole using the radio signal and detects, classifies and controls the pupil. The pupil is detected, classified and controlled by receiving the information of the periphery thereof.

11: Information transmission /
12: Information collecting section
13: Information processor
14: information display section

Claims (2)

An image acquiring unit provided at a front, a side, and a rear end of a pupil of the pupil to acquire front and side topographical images and a ground image on the ground related to the positional information of the duct and the pupil, and a digital camera; An indicator transmission radar installed at a lower portion of the pupil exploration vehicle to acquire pipeline and pupil information; An encoder installed at a front side of the surface transmission radar and measuring a travel distance and a traveling speed of the pupil-searching vehicle when the pupil-surveying vehicle runs on a channel upper surface; And an oscillation frequency acquisition module installed at a lower portion of the pupil-seeking vehicle to acquire a vibration frequency.
An information processing unit incorporating a survey analysis program for collecting and storing the image data acquired by the image acquisition unit and the electromagnetic wave data acquired from the indicator transmission radar and synthesizing, editing, and analyzing the images using the collected data; And
And an information display unit for displaying information processed by the information processing unit,
The information processing unit extracts Mel-Frequency Cepstrum Coefficient (MFCC), Time domain features, and Frequency domain features from the vibration vibration information acquired by the vibration frequency acquisition module, A vibration frequency feature extraction module for extracting a vibration frequency characteristic of the vibration frequency;
A pupil recognition module for detecting and classifying pupils by inputting vibration frequencies acquired in real time from the vibration frequency feature extraction module through a previously learned DNN model as a DNN (Deep Neural Network) which is one of deep learning algorithms.
In order to avoid the over-fitting of the DNN model for the pupil information detected and classified by the pupil recognition module and to avoid the mutual adaptation between the nodes of the neural network model, A pupil detection and classification module for re-detecting and re-classifying the pupils by learning after excluding the pupils;
Wherein the vibration frequency characteristic extracted from the vibration frequency feature extraction module is optimized by using a correlation-based feature selection (CFS) or a convolutional neural network (CNN) Further comprising an optimization module,
A monitoring module installed at a lower side of the manhole cover to detect and communicate the pupil state; A vibration sensor installed in the monitoring module to generate vibration information; An antenna unit for communication; A manhole cover, and a control unit for collecting information on the manhole and its periphery through the vibration sensor unit and transmitting manhole information including vibration information and information on the periphery thereof to an information processing unit through an antenna unit, Wherein the information processing unit connected to the information processing unit receives the unique number of the manhole cover and information on the periphery of the manhole cover in a radio signal to detect, classify, and control pupil detection.
delete

Images