CN112415511A

CN112415511A - Method for removing ground waves from ground radar signals by shallow layer based on wavelet transformation

Info

Publication number: CN112415511A
Application number: CN202011410057.0A
Authority: CN
Inventors: 黄敏; 刘天恒; 彭辉; 余世杰; 桂仲成
Original assignee: Anhui Guimu Robot Co Ltd
Current assignee: Anhui Guimu Robot Co Ltd
Priority date: 2020-12-04
Filing date: 2020-12-04
Publication date: 2021-02-26

Abstract

The invention discloses a method for removing ground waves from a ground radar signal by a shallow layer based on wavelet transformation, which comprises the following steps: sampling continuous radar waveforms according to an isochronous window, wherein any radar signal Ascan in any isochronous window contains m acquisition points; the single-channel radar generates n radar signals Ascan during operation and is marked as channel radar waves bscan; carrying out continuous wavelet transformation on any radar signal Ascan of the channel radar waves bscan by adopting Mexico cap wavelets, and carrying out connected domain searching and screening; searching a position k where any radar signal Ascan in the channel radar waves bscan is closest to the Mexican hat waveform; sequentially arranging the positions K of any radar signal Ascan to obtain a position set sequence K; and smoothing the position set sequence K and removing the ground waves. Through the scheme, the method has the advantages of simple logic, reliable removal, less calculation workload and the like.

Description

Method for removing ground waves from ground radar signals by shallow layer based on wavelet transformation

Technical Field

The invention relates to the technical field of processing of a ground radar signal, in particular to a method for removing ground waves from the ground radar signal in a shallow layer based on wavelet transformation.

Background

Ideally, the shallow ground penetrating radar device is tightly attached to the ground when working, and the collected radar wave data starts from the ground. However, in practical situations, in order to prevent the radar device from being worn, the radar device may involve a certain distance from the ground (as shown in fig. 1), and in an operating state, due to the fact that the radar channel receiving devices are installed on the ground, the ground is not flat, the carrying device per se bumps, and the like, in a slice diagram, each ascan shows different distances from the ground, as shown in fig. 2, fig. 2 is a partial slice of radar data, and it can be seen from the diagram that: each radar ascan data has a truncated section with almost no fluctuation at all. According to experience, the data of the section is considered as the distance from the transmitting device to the ground surface, and the position of each radar Ascan to the ground wave is obviously fluctuated in the upper graph. In this situation, the radar imaging is affected, as well as the distance of the measurement target from the ground.

At present, the methods for removing the ground direct wave in the prior art mainly include the following methods:

firstly, directly removing the direct wave signal, and directly moving the deeper signal data upwards, which is equivalent to changing the effective data volume of an asan; as shown in fig. 3, it is disadvantageous that it is suitable for the stationary state of the ground penetrating radar, and the distance changes at any time in the continuous motion state, so that the length of the direct wave of the ground penetrating radar is obviously not suitable for being fluctuated by adopting the direct removing method.

Secondly, directly removing the direct wave and signal fluctuation generated by the ground, keeping the data volume of each Ascan unchanged, and changing the value of the ground position; as shown in fig. 4, such as "removal of direct arrival of ground penetrating radar signal" by bazedoxifene et al; then, as the Chinese invention patent with the patent publication number of 'CN 107479042A' and the patent name of 'an estimation method of the water storage capacity of the karst zone of the surface layer', the direct starting position is moved to the zero point by analyzing the time of the direct wave at the first wave peak;

third, in the chinese invention patent with patent application No. 202010997862.1 entitled "method for removing ground waves from radar signals based on wavelet transformation" which employs an extremum method to search for the extremum point position of an ascan waveform, denoted as P, within a certain range [0, L ] and to correct the position within a certain range, to search for a new position in [ P-a, P + a ] and record the new position as the final found ground wave position, assuming that the new position q is greater than the amplitude of P, the method is effective in general because the initial search range L and the subsequent correction range a are preset empirical values according to the operating state of the device, and when the device operates in an unconventional operating environment, such as a particularly bumpy road, the above ranges may not contain optimal solutions and are prone to local optimal solutions, as shown in fig. 5. In this technique, the point P is the extreme point of the initial search, and q is the point at which the absolute value of the amplitude of the correction is the maximum, but the positive solution is the point R, and the case of finding the peak and the case of finding the trough are the same as in the above diagram. However, with some radar data in a severe working environment, the ground direct wave cannot be removed well by using the current extreme method.

Therefore, a method for removing ground waves from the radar signals in a shallow layer with simple logic and less calculation workload is urgently needed.

Disclosure of Invention

In view of the above problems, the present invention aims to provide a method for removing ground waves from a radar signal in a shallow layer based on wavelet transform, and the technical scheme adopted by the present invention is as follows:

the method for removing ground waves from a ground radar signal based on a shallow layer of wavelet transform is characterized in that the ground radar signal consists of n single-channel radar waveforms and is recorded as Ascan; n is a natural number greater than 0; the method for removing the ground waves of the ground radar signals by the shallow layer comprises the following steps:

sampling continuous radar waveforms according to an isochronous window, wherein any radar signal Ascan in any isochronous window contains m acquisition points; m is a natural number greater than 0;

the single-channel radar generates n radar signals Ascan during operation and is marked as channel radar waves bscan;

carrying out continuous wavelet transformation on any radar signal Ascan of the channel radar waves bscan by adopting Mexico cap wavelets, and carrying out connected domain searching and screening; searching a position k where any radar signal Ascan in the channel radar waves bscan is closest to the Mexican hat waveform;

sequentially arranging the positions K of any radar signal Ascan to obtain a position set sequence K;

and smoothing the position set sequence K and removing the ground waves.

Further, continuous wavelet transformation is carried out on any radar signal Ascan of the channel radar wave bscan by adopting Mexico cap wavelets, and connected domain searching and screening are carried out; the method comprises the following steps:

carrying out continuous wavelet transformation on a certain radar signal Ascan of a channel radar wave bscan by adopting Mexico cap wavelet to obtain a complex matrix C corresponding to the radar signal Ascan;

obtaining a corresponding module value of any one of the plurality of matrixes C, and obtaining a matrix R; the abscissa of the matrix R is the position of Ascan, and the ordinate of the matrix R is the scale of the wavelet;

normalizing the matrix R to be 0-255 to obtain a matrix norm _ R;

carrying out binarization processing on the matrix norm _ R, and presetting a threshold value to obtain a binarization image norm _ R _ BW larger than the threshold value;

performing connected domain search on the binarized image norm _ R _ BW, and gathering adjacent pixels into the same region to obtain an image norm _ R _ BW _ connect;

and screening the connected domain of the image norm _ R _ BW _ connect, and selecting the position geometry corresponding to the pixel contained in the connected domain with the largest area to obtain an image allLocP.

Further, the step of finding the position k where any radar signal Ascan in the channel radar wave bscan is closest to the mexican hat waveform comprises the following steps:

acquiring any position of a complex matrix C, wherein the real part of the complex matrix C is C1 r; the positive and negative of the real part c1r represent the current position and mexican hat waveform direction relation;

sequentially extracting a set of values of the position pi in the image allLocP corresponding to the complex matrix C, and obtaining a complexSet;

sequentially extracting values of positions pi in the image allLocP, which correspond to the matrix R, and marking the values as MagSet;

obtaining a position P which is positive and corresponds to the MagSet which is the maximum value of the matrix R and the real part of the complex number of the set complexSet; the abscissa of the position P is the position k where the radar signal Ascan is closest to the mexican hat waveform.

Preferably, the smoothing the position set sequence K and removing the ground waves includes the following steps:

cutting the position set sequence K into a plurality of segments and marking the segments as sub _ K;

finding the value V in any segment sub _ k_iAnd a value V_iThe number of occurrences of (c);

any value V in any segment sub _ k is traversed_i：

If V_i-flgV.ltoreq.2 or flgV-V_iIf V is less than or equal to 2, then V is added_iThe value of (a) is set to flgV, and a smoothed sub _ k sequence is obtained; the flgV represents a value V in any segment_iThe maximum number of occurrences of;

and removing signals from the signal to the calculated ground wave position in any radar signal Ascan according to the smoothed position set sequence K, thereby finishing the ground wave removing work.

Furthermore, the smoothing the position set sequence K and removing the ground waves includes the following steps:

obtaining a first derivative DiffK of the position set sequence K;

sequentially taking a subsequence bDiffK from the initial position of the position set sequence K by adopting a sliding window; the length of the subsequence subDiffK is a natural number greater than 1:

if subDiffK [0] is 0, the sub-sequence is not checked, and the next sub-sequence is taken for checking; the subiffk [0] represents a first derivative of the starting position of the position set sequence K;

if subiffk [0] ═ Val, then a position point u is queried in the position set sequence K, and its subiffk [ u ] ═ Val; val is a number different from 0; the value of u is less than the length of the subsequence subDiffK; assigning sequences corresponding to subidifk [1] to subidifk [ u ] as sequences when subidifk [0] is equal to 0: the ground wave removing work is completed by removing signals from the signal opening to the calculated ground wave position in any radar signal Ascan.

Compared with the prior art, the invention has the following beneficial effects:

(1) the method adopts the combination of Mexico hat wavelet function and the mode of directly removing the ground waves, and has simple logic, less calculation workload and accurate removal;

(2) according to the method, the problem that more response areas are formed in the Mexico cap waveform is solved by setting a binarization threshold value and searching and screening the connected domain, so that the calculation workload of the position k where the radar signal Ascan is closest to the Mexico cap waveform is reduced;

(3) the method carries out smoothing processing on the set sequence, judges and removes the ground waves by using a first-order derivative mode, and has the advantages that whether the unsmooth part is a real situation or not is judged through the derivative, and if the unsmooth part is a false unsmooth part, the unsmooth part is smoothed, so that the K sequence which is calculated at first is more reliable.

(4) The method comprises the steps of performing wavelet decomposition on the ascan data to find the accurate position of the Mexico cap, correcting the ground wave positions of all ascans of the whole bscan by combining an extreme method, and finally removing the data above the ground wave position corresponding to each ascan, namely removing the ground direct wave.

In conclusion, the method has the advantages of simple logic, reliable removal, less calculation workload and the like, and has high practical value and popularization value in the technical field of radar signal processing.

Drawings

In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings needed to be used in the embodiments will be briefly described below, it should be understood that the following drawings only illustrate some embodiments of the present invention, and therefore should not be considered as limiting the scope of protection, and it is obvious for those skilled in the art that other related drawings can be obtained according to these drawings without inventive efforts.

Fig. 1 is a diagram illustrating a radar data collection operation in the prior art.

Fig. 2 is a slice of radar data under a sliding window in the prior art.

Fig. 3 is a slice view of fig. 2 with the radar data directly removed.

Fig. 4 is a slice of fig. 2 using radar data after changing the ground position.

Fig. 5 is a schematic diagram of an extremum processing method in the prior art.

FIG. 6 is a diagram of radar data prior to removal of the direct ground wave in accordance with the present invention.

FIG. 7 is a diagram of radar data after removal of the direct ground wave in accordance with the present invention.

FIG. 8 is a norm _ R diagram in the present invention.

FIG. 9 is a connected component search diagram of the present invention.

FIG. 10 is a diagram of connected component search and screening in accordance with the present invention.

FIG. 11 is a waveform (in the same direction) of selected Mexican caps in accordance with the present invention.

Fig. 12 is a waveform (inverted) of selected mexican caps in accordance with the present invention.

Detailed Description

To further clarify the objects, technical solutions and advantages of the present application, the present invention will be further described with reference to the accompanying drawings and examples, and embodiments of the present invention include, but are not limited to, the following examples. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present application.

Examples

As shown in fig. 6 to 12, the present embodiment provides a method for removing ground waves from a radar signal based on a shallow layer of wavelet transform, which includes the following steps:

in the first step, in the whole radar data, the radar data is composed of n single-channel radar waveforms, which are recorded as radar data asan, and in the digital signal, continuous radar waveforms are sampled according to an isochronous window, each asan has m sampling points, and in general, m is 256 or 512. Wherein, the value of k at a certain position of the radar data Ascan is recorded as Ascan [ k ], where k is more than or equal to 0 and less than or equal to m, each value of the radar data Ascan is in a range determined by radar equipment, in the embodiment, 32768< Ascan [ k ] < 32767;

secondly, generating n Ascan data by a single-channel radar in the operation process, recording the n Ascan data as a bscan, wherein the ground wave removal is mainly due to the distance measurement of a ground-bottom target and the display of radar data, but the single radar data are generally displayed less, and one ground-bottom detection ground bottom is composed of a plurality of Ascans, so the ground direct wave removal processing of the shallow radar is generally considered to be the information processing of the bscan;

thirdly, performing continuous wavelet transform on one Ascan in the bscan by using Mexico hat wavelet to obtain a wavelet coefficient which is a complex matrix C, calculating a module value of each complex number in the matrix to obtain a matrix R, wherein the abscissa of the matrix R represents the position of the Ascan, the ordinate represents the scale of the wavelet, the value of the matrix R is normalized to 0-255 and is marked as norm _ R, the data precision of the matrix R is used for participating in subsequent operation, and the norm _ R is directly displayed according to pictures for explaining the embodiment.

The wavelet transformation is carried out by using the Mexican hat wavelet function, a series of parameters are required to be set according to the acquisition frequency of the used radar equipment, and the wavelet scale is required to be transformed, but the wavelet transformation is directly used without any change in the known basic method, so that the wavelet transformation is not described

Fourthly, illustrating that norm _ R is an ideal case, but there are still few cases and many response regions of mexican hat waveforms, so that screening needs to be performed in these response regions, and binarization processing is performed on norm _ R, and a threshold value for binarization may be selected according to actual conditions, as shown in fig. 9, in this embodiment, a maximum pixel value 0.4 is selected as a threshold value, and a binarization graph is marked as norm _ R _ BW.

And step five, as shown in fig. 10, performing connected domain search on the binarized map norm _ R _ BW, and marking as norm _ R _ BW _ connect, and grouping adjacent pixels into the same region.

Sixthly, screening connected domains in norm _ R _ BW _ connect, and selecting the position geometry corresponding to the pixel contained in the selected connected domain with the largest area as allLocP;

seventhly, obtaining a value of any position in the original coefficient matrix C, where the value is a complex number and is denoted as C1, a real part of C1 is denoted as C1r, and the positive and negative of C1r indicate whether the current position and the selected mexican cap waveform are in the same direction, in this embodiment, the selected mexican cap waveform needs C1r to be positive.

Eighthly, sequentially extracting a set of values of the position pi in the image allLocP corresponding to the complex matrix C, wherein the set is complexSet;

the ninth step, sequentially extracting values of positions pi in the image allLocP corresponding to the matrix R, and marking the values as MagSet;

step ten, obtaining a position P which is the maximum value of the matrix R of the MagSet and is positively corresponding to the real part of the complex number of the complexSet set; the abscissa of the position P is the position k where the radar signal Ascan is closest to the mexican hat waveform.

And a tenth step, performing the third step to the tenth step on each Ascan of the whole bscan, and calculating to obtain a ground wave position sequence of each Ascan, wherein K is [ K1, K2, K3 … … … kn ], n is the number of ascans in the bscan, and n is the number of radar data ascans in the bscan.

The eleventh step, performing smoothing processing on the position set sequence K, and removing the ground waves, which provides two ways in this embodiment:

the first method comprises the following steps:

(1) in the present embodiment, every 2000K is recorded as a segment, sub _ K, and the segment length may be changed according to the actual situation.

(2) In each sub _ k, calculate how many different values there are in total, and how many positions there are for each value, e.g. a 2000 number in sub _ k, remove duplicates, only have 4 values v0, v1, v2, v3, where v0 has 1200, take v0 of the largest number of hits as flgV of the current sub _ k,

(3) traversing each value v in sub _ K, if v-flgV < ═ 2 or flgV-v < ═ 2, the current value v is set to flgV.

(4) And (4) repeating the steps (1) to (3), wherein all sub _ Ks are smoothed to form a smoothed K sequence, the K sequence is the ground wave position corresponding to each Ascan in the current bscan, and signals from the signal opening to the calculated ground wave position in any radar signal Ascan are removed according to the ground wave position corresponding to each Ascan in the K sequence, so that the ground wave removal work is completed.

And the second method comprises the following steps:

(1) calculating a first derivative DiffK of the sequence K, wherein DiffK [ i ] ═ K [ i +1] -K [ i ]; wherein DiffK represents the degree of mutation at the ground wave position;

(2) starting from position 0 in the first-order derivative DiffK, sequentially taking a subsequence with the length set as bumpyGap by using a sliding window, wherein in the subsequence with the length, whether the detection is sudden change or not is detected, the bumpyGap is 3, and the subsequence is marked as SubDiffK;

(3) if subDiffK [0] is 0, the subsequence is not checked, and the next subsequence is taken for checking;

if subiffk [0] is Val, finding a position in the subsequence, subiffK [ u ] is Val, u < bumpyGap, which indicates that one/a plurality of ground wave positions exist in the bumpyGap range, jumping occurs due to unknown reasons, and the normal state is recovered in a certain range, and indicates that the area is a ground wave position abnormal area and needs to be repaired, correcting the position in the subiffK corresponding to the sequence K, and setting the ground position of the abnormal section as a K value corresponding to subiffK [0 ].

At this time, the sequence K is the ground wave position corresponding to each Ascan in Bscan, and the signals from the signal start segment to the calculated ground wave position in each Ascan are removed, so that the ground wave removal for Bscan is completed.

The above-mentioned embodiments are only preferred embodiments of the present invention, and do not limit the scope of the present invention, but all the modifications made by the principles of the present invention and the non-inventive efforts based on the above-mentioned embodiments shall fall within the scope of the present invention.

Claims

1. The method for removing ground waves from a ground radar signal based on a shallow layer of wavelet transform is characterized in that the ground radar signal consists of n single-channel radar waveforms and is recorded as Ascan; n is a natural number greater than 0; the method for removing the ground waves of the ground radar signals by the shallow layer comprises the following steps:

and smoothing the position set sequence K and removing the ground waves.

2. The method for removing ground waves from a ground radar signal based on the shallow layer of wavelet transform as recited in claim 1, wherein continuous wavelet transform is performed on any radar signal Ascan of a channel radar wave bscan by using Mexico hat wavelet, and connected domain search and screening are performed; the method comprises the following steps:

normalizing the matrix R to be 0-255 to obtain a matrix norm _ R;

3. The wavelet transform-based method for removing ground waves from radar signals in shallow layers, according to claim 2, wherein a position k where any radar signal Ascan in a channel radar wave bscan is closest to a Mexico hat waveform is found, comprising the following steps:

4. The wavelet transform based method for removing ground waves from radar signals in shallow layers according to claim 1, 2 or 3, wherein the smoothing process is performed on the position set sequence K, and ground waves are removed, comprising the following steps:

any value V in any segment sub _ k is traversed_i：

5. The wavelet transform based method for removing ground waves from radar signals in shallow layers according to claim 1, 2 or 3, wherein the smoothing process is performed on the position set sequence K, and ground waves are removed, comprising the following steps:

obtaining a first derivative DiffK of the position set sequence K;