A method for correcting range effect of hyperspectral lidar based on homogeneous targets

Abstract

The present invention discloses a method for correcting the distance effect of a hyperspectral laser radar based on a homogeneous target. First, a hyperspectral laser radar is used to measure the original echo intensity data of a reference homogeneous target at different distances at all wavelengths; the measured original echo intensity data is normalized to obtain the normalized dimensionless echo intensity data; a segmented fitting model method is used to calculate the distance effect function; and according to the constructed distance effect function, a certain standard distance R is used to correct the distance effect function. s The distance effect correction factor is calculated as a reference; the original echo intensity data at any distance R is corrected according to the calculated distance effect correction factor to obtain the echo intensity data after distance effect correction. The above method fully considers the identity of the distance effect under different hyperspectral lidar wavelengths and is not limited by the vegetation index method. It can obtain the echo intensity after distance effect correction at any distance, laying the foundation for the subsequent vertical estimation of biochemical parameters.

Description

Technical Field

[0001] The present invention relates to the technical field of hyperspectral laser radar, and in particular to a method for correcting distance effect of a hyperspectral laser radar based on homogeneous targets. Background Art

[0002] The acquisition of hyperspectral lidar point cloud data mainly relies on two methods. One is based on the combination of multiple single-band lidar systems, and the other is the fusion of traditional single-band lidar and hyperspectral imager data. However, the former has data matching and calibration problems caused by the different positions of laser light sources of multiple sensors, and the latter has the problem of difficult matching and large errors when fusing two-dimensional hyperspectral pixels with three-dimensional point cloud data. The emergence of hyperspectral lidar has broken through the technical bottleneck of the aforementioned methods. Hyperspectral lidar can be used to directly obtain hyperspectral three-dimensional point cloud data of the target object. However, hyperspectral lidar also faces the problem of distance effect. For the same target object, the echo intensity collected by the hyperspectral lidar will first increase gradually with the increase of distance, and then gradually decrease. It is not fixed, which will bring certain errors when inverting the three-dimensional biochemical parameters of vegetation, greatly affecting the ability of hyperspectral lidar to estimate the physical and chemical properties of the target.

[0003] At present, the methods for correcting the distance effect of traditional single-band lidar are mainly divided into theoretical models, polynomial models and vegetation index methods. Among them, the theoretical model is derived from the principle of laser radar radiation transmission, and the distance effect of single-band laser radar is characterized by the inverse square method of distance, that is, for the same target, theoretically, the echo intensity received by the laser radar is inversely proportional to the square of the distance, and the echo intensity gradually decreases with increasing distance. The disadvantage of this method is that it ignores the close-range defocusing effect of the laser radar. The laser radar system will adjust the echo signal, reduce the strong signal at close range, and amplify the weak signal at long distance. Therefore, the distance effect does not follow the theoretical model; the polynomial model regards the influence of reflectivity and incident angle effects as a fixed value K within a certain measurement distance range, and fits the change of single-band echo intensity and distance with a polynomial function, and uses the fitted polynomial function to correct the echo intensity corresponding to different distances. However, this method is aimed at single-band lidar. Since hyperspectral lidar introduces more spectral information, more wavelengths have to be taken into account when eliminating the distance effect, and the distance effect phenomenon and correction function of different wavelengths need to be re-studied; the vegetation index rule does not correct the echo intensity at different distances at all, and only uses ratios or other construction methods to construct the vegetation index. As a result, a vegetation index value is obtained. The disadvantage is that users cannot directly obtain the echo intensity information after the distance effect of all wavelengths is corrected. If the vegetation index is not suitable for subsequent quantitative research, it cannot be used.

[0004] Therefore, it is urgent to study the distance effect phenomenon of hyperspectral lidar and find a method suitable for correcting the distance effect of hyperspectral lidar. Summary of the invention

[0005] The purpose of the present invention is to provide a method for correcting the distance effect of a hyperspectral lidar based on a homogeneous target. The method fully considers the identity of the distance effect under different hyperspectral lidar wavelengths and is not limited by the vegetation index method. The method can obtain the echo intensity after the distance effect is corrected at any distance, laying a foundation for the subsequent vertical estimation of biochemical parameters.

[0006] The objective of the present invention is achieved through the following technical solutions:

[0007] A method for correcting range effect of a hyperspectral laser radar based on homogeneous targets, the method comprising:

[0008] Step 1: Use a hyperspectral laser radar to measure the raw echo intensity data of a reference homogeneous target at different distances at all wavelengths; wherein the reference homogeneous target is a 99% standard reflectivity plate;

[0009] Step 2, normalizing the original echo intensity data measured in step 1 to obtain normalized dimensionless echo intensity data;

[0010] Step 3: for the dimensionless echo intensity data after normalization in step 2, the distance effect function is calculated by using a piecewise fitting model method;

[0011] Step 4: Based on the distance effect function constructed in step 3, a standard distance R s For reference, calculate the distance effect correction factor;

[0012] Step 5: According to the distance effect correction factor calculated in step 4, the original echo intensity data at any distance R is corrected to obtain the hyperspectral lidar echo intensity data after distance effect correction.

[0013] It can be seen from the technical solution provided by the present invention that the above method fully considers the identity of the distance effect under different hyperspectral lidar wavelengths, and is not limited by the vegetation index method. It can obtain the echo intensity after distance effect correction at any distance, laying the foundation for the subsequent vertical estimation of biochemical parameters. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0015] Figure 1 A schematic flow chart of a method for correcting range effect of a hyperspectral laser radar based on homogeneous targets provided in an embodiment of the present invention;

[0016] Figure 2 This is a schematic diagram of the variation of the original echo intensity of the 99% standard reflectivity plate as a function of distance in the example of the present invention;

[0017] Figure 3 This is a schematic diagram of the change of echo intensity with distance after normalization of a 99% standard reflectivity plate according to an example of the present invention;

[0018] Figure 4 It is a schematic diagram of the echo intensity of a standard reflector after the distance effect correction is performed using the inverse square method of distance, the polynomial model method and the method described in the embodiment of the present invention;

[0019] Figure 5 It is a schematic diagram of error evaluation of distance effect correction of a 99% standard reflectivity plate using the distance inverse square method and the polynomial model method and the method described in the embodiment of the present invention;

[0020] Figure 6 This is a schematic diagram of the variation of the original echo intensity of the vegetation leaf sample as a function of distance in the example of the present invention;

[0021] Figure 7 This is a schematic diagram of the variation of echo intensity with distance after normalization of vegetation leaf samples in the example of the present invention;

[0022] Figure 8 It is a schematic diagram of the echo intensity of vegetation leaf samples after the distance effect correction is performed using the inverse distance square method and the polynomial model method and the method described in the embodiment of the present invention;

[0023] Fig. 9 It is a schematic diagram of error evaluation for correcting distance effect of blade samples using the inverse distance square method, the polynomial model method and the method described in the embodiment of the present invention. DETAILED DESCRIPTION

[0024] The following is a clear and complete description of the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments, which does not constitute a limitation of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0025] like Figure 1 The figure is a schematic flow chart of a method for correcting the range effect of a hyperspectral laser radar based on homogeneous targets provided by an embodiment of the present invention, the method comprising:

[0026] Step 1: Use a hyperspectral laser radar to measure the raw echo intensity data of a reference homogeneous target at different distances at all wavelengths; wherein the reference homogeneous target is a 99% standard reflectivity plate;

[0027] In this step, a 99% standard reflectivity plate is selected as a reference homogeneous target for distance effect correction. The reason is that only the system effect of the lidar itself needs to be considered during the distance effect measurement process. In the process of using hyperspectral lidar for measurement, as the distance increases, the surface reflectivity of the target object needs to remain unchanged;

[0028] Among them, according to the maximum measurement range of the hyperspectral laser radar, the distance change is set to 2.5m-22m, with a step length of 0.5m. After each measurement, the distance between the reference homogeneous target and the hyperspectral laser radar increases by 0.5m;

[0029] The wavelength range of hyperspectral lidar is 523nm-800nm, with a total of 16 bands, such as Figure 2 The figure shows the variation of the original echo intensity of the 99% standard reflectivity plate with distance according to the example of the present invention.

[0030] In a specific implementation, the reason why this embodiment does not set a longer distance is that experiments have found that a too long distance will lead to unstable received signals and weak echo intensity.

[0031] Step 2, normalizing the original echo intensity data measured in step 1 to obtain normalized dimensionless echo intensity data;

[0032] In this step, the specific calculation formula for normalization is:

[0033]

[0034] Among them, I stan (R) is the normalized intensity value corresponding to the 99% standard reflectivity plate at a certain distance R; I raw (R) is the echo intensity value obtained by the original measurement at the distance R; max is the maximum original echo intensity at the corresponding wavelength.

[0035] For example, for the distance experimental data of each band of the 99% standard reflectivity plate collected by the hyperspectral lidar, the maximum original echo intensity I corresponding to the first band 523nm is calculated first. max , and then calculate the original echo intensity value I at each distance R in this band (523nm) raw (R) is normalized by division to obtain the normalized echo intensity I of this band (523nm) stan (R), the value range is 0-1.

[0036] The above calculation is performed for all bands to obtain the normalized dimensionless echo intensity data of all bands, such as Figure 3 It is a schematic diagram showing the variation of echo intensity with distance after normalization processing of the 99% standard reflectivity plate in the example of the present invention.

[0037] Step 3: for the dimensionless echo intensity data after normalization in step 2, the distance effect function is calculated by using a piecewise fitting model method;

[0038] In this step, the specific fitting formula is as follows:

[0039] f(R)=aR 2 +bR+c,R≤R seg ;

[0040] f(R)=de -fR +g,R>R seg

[0041] Where R is the distance from the hyperspectral laser radar to the target object; f(R) is the value of the distance effect function; a, b, c, d, f, g are the parameters to be estimated by the model, which are obtained by fitting using the least squares method; R seg is the distance segmentation value, indicating that the normalized echo intensity corresponding to the distance at this point is used as the dividing value. The first part of the fitting formula adopts a quadratic polynomial fitting model, and the second part adopts an exponential model.

[0042] For example, for the normalized dimensionless echo intensity data, 6.5m is used as the distance segmentation value R seg The normalized intensity-distance phenomenon less than or equal to 6.5m is fitted by a quadratic function, and the normalized intensity-distance phenomenon greater than 6.5m is fitted by an exponential function based on the effect of the laser radar that gradually decreases with increasing distance.

[0043] Substituting all the distances R of the corresponding band and the normalized distance effect function value f(R) into the above formula, the least square method is used for fitting, and the parameter values ​​of a, b, c, d, f, and g can be estimated. It should be noted that the number of decimal places retained in these parameter values ​​will affect the accuracy of the distance effect function. The parameter results obtained in the example of the present invention are shown in Table 1 below:

[0044] Table 1 Hyperspectral LiDAR distance effect function parameter table

[0045] a b c d f g -0.07868 0.85736 -1.33055 2.67067 0.16357 0.16357

[0046] Step 4: Based on the distance effect function constructed in step 3, a standard distance R s For reference, calculate the distance effect correction factor;

[0047] In this step, the distance effect correction factor is calculated as follows:

[0048]

[0049] Where i(R) is the distance effect correction factor; f(R s ) is the standard distance R s The distance effect function value at is the segmented model function value obtained by fitting in step 3; f(R) is the distance effect function value corresponding to any distance.

[0050] Specifically, if the experiment obtains a scanning point cloud of a certain scene, in order to eliminate the distance effect of the point cloud in the scene, it is necessary to calibrate the echo intensity at different distances. Generally, the average distance between the scene target and the laser radar can be used as the standard distance R s , and correct the echo intensity at other distances to this distance.

[0051] For example, in this embodiment, 8.7 m is used as the standard distance R s The process of distance effect correction is: first calculate the wavelength 523nm and the standard distance R s =8.7m distance effect function value; then calculate the wavelength 523nm and distance R = 2.5m distance effect function value, the ratio of the two is recorded as the correction factor i (R).

[0052] Step 5: According to the distance effect correction factor calculated in step 4, the original echo intensity data at any distance R is corrected to obtain the hyperspectral lidar echo intensity data after distance effect correction.

[0053] In this step, the calculation formula for correcting the original echo intensity data at any distance R is as follows:

[0054] I cor (R)=I raw (R)×i(R)

[0055] Among them, I cor (R) is the corrected echo intensity at distance R; i(R) is the distance effect correction factor; I raw (R) is the echo intensity value obtained by the original measurement at the distance R;

[0056] By multiplying the original echo intensity value at a certain wavelength and a certain distance by the distance effect correction factor, the corrected echo intensity value at the wavelength and the distance can be obtained; the calculation process of steps 4 and 5 is performed for all distances to obtain the corrected echo intensity values ​​at all distances under the corresponding wavelength, and then traversing all wavelengths in turn to obtain the corrected echo intensity values ​​at all distances under all wavelengths.

[0057] For example, by multiplying the original echo intensity value at a wavelength of 523nm and a distance of R=2.5m by the correction factor, the corrected echo intensity value at R=2.5m can be obtained; performing the calculation process of steps 4 and 5 for all distances can obtain the corrected echo intensity values ​​corresponding to all distances at the wavelength; then traversing all wavelengths in turn and executing the operations of steps 4 and 5 can obtain the corrected echo intensities at all distances at other wavelengths.

[0058] The effect and error of the correction method described in the embodiment of the present invention are analyzed as follows:

[0059] like Figure 2 The figure shows the variation of the original echo intensity of the 99% standard reflectivity plate with distance in the example of the present invention. It can be seen that the echo intensity of the hyperspectral laser radar does not always decrease gradually with increasing distance. There is an obvious abnormal phenomenon at a close distance of 2.5m-5.5m, and the echo intensity gradually increases with increasing distance. This is caused by the defocusing effect of the system.

[0060] like Figure 3 The figure shows the variation of echo intensity with distance after normalization of the 99% standard reflectivity plate in the example of the present invention. Obviously, for each wavelength, the normalized echo intensity shows obvious identity of distance effect. Therefore, the distance effect of the hyperspectral lidar is independent of the wavelength. For different wavelengths, a unified distance effect function can be used to characterize its variation characteristics.

[0061] like Figure 4 FIG. 1 is a schematic diagram of the echo intensity of a standard reflector after distance effect correction using the inverse square method and the polynomial model method and the method described in the embodiment of the present invention, including the inverse square method ( Figure 4 (a)), polynomial model method ( Figure 4 (b)), the method described in the embodiment of the present invention ( Figure 4 (c) Schematic diagram of the echo intensity of three 99% standard reflectivity plates after correction. Figure 2 Compared with the original echo intensity, the echo intensity after distance inverse square correction ( Figure 4 (a)) performs well when the distance effect is greater than 6.5m, and no obvious fluctuation is found. However, the echo intensity after distance effect correction within the defocusing effect range changes greatly from 0 to 15mV. Therefore, this method is not suitable for the pre-focus distance effect correction process of hyperspectral lidar. For the echo intensity corrected by the polynomial model method ( Figure 4 (b)) and the echo intensity corrected by the segmented fitting model method ( Figure 4(c)) Both of them have a good effect on correcting the original echo intensity, and can correct the echo intensity at other distances to the standard distance. Although the intensity change trend does not seem to be much different, the segmented fitting model method of the present invention has smaller fluctuations in some local areas such as after a larger distance of 20m, which is better than the polynomial model method.

[0062] The above results show that the correction method based on the segmented fitting model of homogeneous targets has a better correction effect on the original echo intensity, and the distance effect has been eliminated overall.

[0063] In addition, in order to quantitatively compare and analyze the correction effect, error analysis was performed. The standard deviation and coefficient of variation of the 99% standard reflectivity plate calibration by the classical method and the segmented fitting model method were calculated respectively. The standard deviation is used to represent the internal fluctuation of the data, and the coefficient of variation is the ratio of the standard deviation to the mean value. By comparing the coefficient with the mean value, the coefficient of variation can more accurately describe the internal fluctuation difference of the data, such as Figure 5 It is a schematic diagram showing error evaluation of distance effect correction of a 99% standard reflectivity plate using the distance square inverse method, the polynomial model method and the method described in the embodiment of the present invention.

[0064] from Figure 5 (a) The standard deviation diagram shows that the original echo intensity changes dramatically and has large internal differences. The internal fluctuation of the corrected echo intensity is smaller than the original echo intensity. Specifically, the segmented fitting model method has the smallest standard deviation and is slightly better than the polynomial model method. The overall standard deviation of the two methods is smaller than the inverse square distance method. Figure 5 (b) Schematic diagram of coefficient of variation It can be seen that the coefficient of variation of the correction method described in the embodiment of the present invention is the smallest, and is smaller than the polynomial model method in many bands, which shows the superiority of the correction method described in the embodiment of the present invention. More importantly, the exponential model is used in the second half of the correction method described in the embodiment of the present invention. At a longer distance than the measurement distance, the operation principle is also better than the polynomial model. The reason is that the polynomial model can only simulate the change function of the measurable data obtained by the experiment. At a longer distance, if fitting is performed, the result may be that the echo intensity rises to a large value again or increases, then decreases, and then increases in a cycle, which does not conform to the phenomenon that the intensity of the laser radar gradually decreases with distance outside the focus.

[0065] In addition, in order to illustrate the effectiveness of the method proposed in the embodiment of the present invention, vegetation leaf samples are used as measurement objects for verification. Figure 6 This is a schematic diagram of the variation of the original echo intensity of the vegetation leaf sample with distance in the example of the present invention. Figure 7 This is a schematic diagram of the variation of echo intensity with distance after normalization of the vegetation leaf sample in the example of the present invention. There is a little difference between different wavelengths at a larger distance, but overall it can still be explained that the distance effect is independent of the wavelength.

[0066] like Figure 8 FIG. 1 is a schematic diagram of the echo intensity of vegetation leaf samples after distance effect correction using the inverse square method and the polynomial model method and the method described in the embodiment of the present invention, including the inverse square method ( Figure 8 (a)) and the polynomial model method ( Figure 8 (b)) The blade sample echo intensity after correction, the correction method of the embodiment of the present invention ( Figure 8 (c)) still shows a certain degree of applicability; by calculating the standard deviation and coefficient of variation, such as Fig. 9 The figure shows a schematic diagram of error evaluation for distance effect correction of blade samples using the inverse distance square method, the polynomial model method, and the method described in the embodiment of the present invention. It can be seen that the standard deviation and coefficient of variation of the correction method described in the embodiment of the present invention are smaller than those of other methods, and the correction effect is better.

[0067] It is worth noting that the contents not described in detail in the embodiments of the present invention belong to the prior art known to professional and technical personnel in the field.

[0068] In summary, the correction method described in the embodiment of the present invention uses a homogeneous target (99% standard reflectivity plate) as the measurement object to estimate the distance effect function of the hyperspectral lidar, which can avoid the reflectivity change error introduced by using natural targets such as vegetation leaves as the measurement objects during the increase in distance, thereby improving the accuracy of the distance effect function. During use, the correction method of the present invention is not limited by the vegetation index method, and the corrected echo intensity at any wavelength and distance can be obtained, which lays the foundation for the subsequent vertical estimation of biochemical parameters.

[0069] At the same time, this method is not limited to a certain wavelength, but is aimed at the echo intensity correction process at different distances under all hyperspectral lidar wavelengths. The reason is that the distance effect of different wavelengths is found to be identical in specific instance measurements. The exponential function used in the second half of the segmented fitting model can still be used in engineering measurement scenarios after exceeding the maximum experimental distance, compared to the quadratic polynomial fitting function. The function value gradually decreases, which is more in line with the lidar distance effect phenomenon. The usual quadratic polynomial fitting function will show a rapid upward trend after exceeding the maximum experimental distance, and can only fit and correct the echo intensity at the distance set during the experiment.

[0070] The above is only a preferred specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by any technician familiar with the technical field within the technical scope disclosed in the present invention should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims. The information disclosed in the background technology section of this article is only intended to deepen the understanding of the overall background technology of the present invention, and should not be regarded as an admission or in any form that the information constitutes prior art known to those skilled in the art.

Claims

1. A method for correcting the range effect of a hyperspectral laser radar based on homogeneous targets, characterized in that: The method comprises: Step 1: Use a hyperspectral laser radar to measure the raw echo intensity data of a reference homogeneous target at different distances at all wavelengths; wherein the reference homogeneous target is a 99% standard reflectivity plate; In step 1, during the measurement using the hyperspectral lidar, the reflectivity of the target surface needs to remain constant as the distance increases; Among them, according to the maximum measurement range of the hyperspectral laser radar, the distance change is set to 2.5m-22m, with a step length of 0.5m. After each measurement, the distance between the reference homogeneous target and the hyperspectral laser radar increases by 0.5m; The wavelength range of hyperspectral lidar is 523nm-800nm, with a total of 16 bands; Step 2, normalizing the original echo intensity data measured in step 1 to obtain normalized dimensionless echo intensity data; In step 2, the specific calculation formula for normalization is: Among them, I stan (R) is the normalized intensity value corresponding to the 99% standard reflectivity plate at a certain distance R; I raw (R) is the echo intensity value obtained by the original measurement at the distance R; max is the maximum original echo intensity at the corresponding wavelength; Step 3: for the dimensionless echo intensity data after normalization in step 2, the distance effect function is calculated by using a piecewise fitting model method; In step 3, the specific fitting formula is as follows: f(R)=aR 2 +bR+c,R≤R seg ; f(R)=the -fR +g,R>R seg Where R is the distance from the hyperspectral laser radar to the target; f(R) is the distance effect function value; a, b, c, d, f, g are the parameters to be estimated by the model, which are obtained by fitting using the least squares method; R seg is the distance segmentation value, which means that the normalized echo intensity corresponding to the distance at this point is used as the dividing value. The first part of the fitting formula adopts a quadratic polynomial fitting model, and the second part adopts an exponential model. Step 4: Based on the distance effect function constructed in step 3, a standard distance R s For reference, calculate the distance effect correction factor; In step 4, the distance effect correction factor is calculated as follows: Where i(R) is the distance effect correction factor; f(R s ) is the standard distance R s The distance effect function value at is the segmented model function value obtained by fitting in step 3; f(R) is the distance effect function value corresponding to any distance; Step 5: Correct the original echo intensity data at any distance R according to the distance effect correction factor calculated in step 4 to obtain the hyperspectral lidar echo intensity data after distance effect correction; In step 5, the calculation formula for correcting the original echo intensity data at any distance R is as follows: I cor (R)=I raw (R)×i(R) Among them, I cor (R) is the corrected echo intensity at distance R; i(R) is the distance effect correction factor; I raw (R) is the echo intensity value obtained by the original measurement at the distance R; By multiplying the original echo intensity value at a certain wavelength and a certain distance by the distance effect correction factor, the corrected echo intensity value at the wavelength and the distance can be obtained; the calculation process of steps 4 and 5 is performed for all distances to obtain the corrected echo intensity values ​​at all distances under the corresponding wavelength, and then traversing all wavelengths in turn to obtain the corrected echo intensity values ​​at all distances under all wavelengths.

Images