Intelligent classification method of diseased and insect-infested trees and dead trees based on multispectral images

Abstract

The present invention discloses an intelligent classification method for pest-infested and dead trees based on multispectral imagery. This method belongs to the field of image processing technology and includes the following steps: S1. Acquiring a multispectral image of the area to be classified, performing geometric and radiometric correction on the multispectral image, and then extracting the pixel values of the pixels in each band in the latest multispectral image; S2. Determining nonlinear adjustment values for each band based on the pixel values of the pixels in each band in the latest multispectral image, and generating color entropy for the latest multispectral image; S3. Determining the pest-infested area in the latest multispectral image and, in the remaining area, determining the dead tree area based on the color entropy of the latest multispectral image. This method maintains robustness to spectral linear variations, significantly improving the processing efficiency of large-scale images while ensuring classification accuracy.

Description

Technical Field

[0001] The present invention belongs to the technical field of image processing, and in particular relates to an intelligent classification method for pest-infested trees and dead trees based on multispectral images. Background Art

[0002] With climate change and the intensification of pests and diseases, the number of dead trees in forests is increasing year by year, severely impacting ecosystems and forestry resource management. Traditional methods for monitoring dead trees rely primarily on manual field surveys, which are inefficient, costly, and difficult to cover large areas. In recent years, unmanned aerial vehicle (UAV) remote sensing technology has gradually been applied to forestry monitoring. However, existing technologies often rely on single spectral information and are often used for dead tree monitoring. However, they suffer from the following problems: 1) Single feature limitation: Traditional methods rely on a single vegetation index (such as NDVI), making it difficult to distinguish between pests and diseases (localized lesions) and dead trees (overall death); 2) Inadequate classification accuracy: Existing algorithms have a high misclassification rate for pests and diseases and dead trees with similar spectral characteristics, especially in complex forest environments. Summary of the Invention

[0003] In order to solve the above problems, the present invention proposes an intelligent classification method for diseased and insect-infested trees and dead trees based on multispectral images.

[0004] The technical solution of the present invention is: a method for intelligently classifying pest-infested trees and dead trees based on multispectral images, comprising the following steps:

[0005] S1. Collect multispectral images of the area to be classified, perform geometric correction and radiation correction on the multispectral images, and then extract the pixel values of the pixels in each band in the latest multispectral images;

[0006] S2, determining the nonlinear adjustment value of each band according to the pixel value of each pixel point in the latest multispectral image in each band, and generating color entropy for the latest multispectral image;

[0007] S3. Determine the pest and disease area of the latest multispectral image, and determine the dead tree area in the remaining area based on the color entropy of the latest multispectral image.

[0008] Furthermore, S2 includes the following sub-steps:

[0009] S21. Adding the ratio of the pixel value of the pixel point in the latest multispectral image in the red band to the maximum red band pixel value, the ratio of the pixel value in the blue band to the maximum blue band pixel value, and the ratio of the pixel value in the green band to the maximum green band pixel value to obtain the color ratio of the pixel point;

[0010] S22. Calculate the color entropy of the latest multispectral image based on the color ratios of all pixels.

[0011] The beneficial effect of the above further scheme is: in the present invention, by comparing the pixel values of the red, yellow and green bands with the maximum pixel values of their respective bands and summing them, the color rate of the pixel point is obtained, which can comprehensively reflect the relative intensity of the pixel point in different color channels and provide a comprehensive indicator for further analysis of color characteristics.

[0012] Furthermore, S22 includes the following sub-steps:

[0013] S221, calculating a nonlinear adjustment value for each band based on the lower quartile and upper quartile of the pixel values of all pixels of the latest multispectral image in each band;

[0014] S222, determining an adjustment range of the latest multispectral image according to the nonlinear adjustment value of each band;

[0015] S223, determining the ratio of the number of pixels whose color ratios fall within the adjustment range to the total number of pixels in the latest multispectral image;

[0016] S224 , determining the color entropy of the latest multispectral image according to the ratio of the number of pixels whose color ratios fall within the adjustment range to the total number of pixels in the latest multispectral image.

[0017] The beneficial effects of this further solution are as follows: In this invention, the use of upper and lower quartiles to calculate nonlinear adjustment values avoids interference from extreme values, making the data more robust and adaptable to complex image features. The adjustment range combines the adjustment values for each band with the maximum pixel value to define the effective range of color ratios and screen representative pixels. Color entropy is calculated based on the proportion of pixels whose color ratios fall within the adjustment range. This pixel distribution ratio is converted into a measure of color distribution uncertainty, providing effective features for the classification of pests, diseases, and dead trees.

[0018] Furthermore, in S221, the nonlinear adjustment value of the band The calculation formula is:

[0019] ;

[0020] Where, Indicates the quartiles of the pixel values of all pixels in the latest multispectral image in the band. Indicates the lower quartile of the pixel values of all pixels in the latest multispectral image in the band. Indicates the maximum pixel value of all pixels in the latest multispectral image in the band, Indicates the minimum pixel value of all pixels in the latest multispectral image in the band, Represents an exponent.

[0021] Furthermore, in S222, the expression of the adjustment range of the latest multispectral image is: ,in, Indicates the nonlinear adjustment value of the red band, Indicates the nonlinear adjustment value of the green band, Indicates the nonlinear adjustment value of the blue band, Indicates finding the minimum value, Indicates the maximum pixel value of all pixels in all bands in the latest multispectral image.

[0022] Furthermore, in S224, the color entropy of the latest multispectral image The calculation formula is:

[0023] ;

[0024] Where, represents the logarithmic function, Indicates the ratio of the number of pixels whose color ratio falls within the adjustment range to the total number of pixels in the latest multispectral image.

[0025] Furthermore, S3 includes the following sub-steps:

[0026] S31, taking the pixel points whose normalized red edge index in the latest multispectral image is less than the set threshold as the diseased and insect-infested tree area;

[0027] S32, based on the remaining areas except the pest-infested tree areas in the latest multispectral image, constructing a screening objective function using the color entropy generated by the latest multispectral image;

[0028] S33, determining the optimal target pixel point using the screening objective function;

[0029] S34. Pixels in the remaining area whose Manhattan distance to the optimal target pixel is less than a set threshold are regarded as dead tree areas.

[0030] The beneficial effect of the above-mentioned further scheme is as follows: In the present invention, the Normalized Red Edge Index (NDRE) is a remote sensing indicator used to assess the health and growth status of vegetation. A decrease in the NDRE value (Normalized Red Edge Index) is directly related to the reduction of chlorophyll caused by pests and diseases, and can be used as an effective screening indicator. Pixels below a set threshold are identified as areas of pest-infested trees. This specific indicator can be used to quickly locate areas potentially affected by pests and diseases, providing a basis for further distinguishing dead trees. A screening objective function is constructed based on the remaining area excluding the pest-infested tree area. This function comprehensively considers the relationship between color entropy and the mean value of pixel values in different bands. It can highlight points in the remaining area that have significant color differences from other pixels, providing an effective mathematical model for screening dead tree areas. The screening objective function is used to determine the optimal target pixel point, which can be regarded as a representative pixel point in the remaining area that is most relevant to the characteristics of dead trees. The pixels in the remaining area whose Manhattan distance to the best target pixel is less than the set threshold are regarded as the dead tree area. The range of dead trees is determined by measuring the color distance. Taking into account the color similarity of dead trees, the dead tree area can be identified more accurately, avoiding the uncertainty and randomness of single pixel judgment.

[0031] Furthermore, in S32, the expression of the screening objective function is:

[0032] ;

[0033] Where, represents the color entropy of the latest multispectral image, Indicates the remaining area The average pixel value of a pixel in all bands, represents the best target pixel point determined by screening the objective function, Represents the mean pixel value of all pixels in the remaining area in the red band, Represents the mean pixel value of all pixels in the remaining area in the green band, Represents the mean pixel value of all pixels in the remaining area in the blue band.

[0034] The beneficial effects of the present invention are:

[0035] (1) The present invention effectively eliminates geometric distortion and radiation error of images through geometric correction and radiation correction, ensuring that pixel values truly reflect the spectral characteristics of ground objects; by calculating the nonlinear adjustment value of each band, the dynamic range of the spectral response of the red, green and blue bands is optimized, and the quartile statistics are used to adaptively compress extreme values and stretch intermediate values, highlighting the subtle spectral differences between diseased and insect-infested trees and dead trees in key bands;

[0036] (2) The color entropy index proposed in this paper innovatively integrates multi-band information and quantifies the health status of vegetation by statistically analyzing the uniformity of color rate distribution: based on the normalized red edge index, it can quickly locate the pest and disease areas, and in the remaining areas, it can accurately identify dead trees with homogenized spectral characteristics by combining color entropy thresholds with Manhattan distance clustering;

[0037] (3) The present invention is applicable to vegetation monitoring in different regions and seasons. It has strong generalization ability and significantly improves the sensitivity to changes in the physiological state of vegetation while maintaining robustness to spectral linear changes. While ensuring classification accuracy, it greatly improves the processing efficiency of large-scale images. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 This is a flowchart of the intelligent classification method for diseased and insect-infested trees and dead trees based on multispectral images. DETAILED DESCRIPTION

[0039] The embodiments of the present invention will be further described below with reference to the accompanying drawings.

[0040] like Figure 1 As shown, the present invention provides an intelligent classification method for pest-infested trees and dead trees based on multispectral images, comprising the following steps:

[0041] S1. Collect multispectral images of the area to be classified, perform geometric correction and radiation correction on the multispectral images, and then extract the pixel values of the pixels in each band in the latest multispectral images;

[0042] S2, determining the nonlinear adjustment value of each band according to the pixel value of each pixel point in the latest multispectral image in each band, and generating color entropy for the latest multispectral image;

[0043] S3. Determine the pest and disease area of the latest multispectral image, and determine the dead tree area in the remaining area based on the color entropy of the latest multispectral image.

[0044] In this embodiment of the present invention, S2 includes the following sub-steps:

[0045] S21. Adding the ratio of the pixel value of the pixel point in the latest multispectral image in the red band to the maximum red band pixel value, the ratio of the pixel value in the blue band to the maximum blue band pixel value, and the ratio of the pixel value in the green band to the maximum green band pixel value to obtain the color ratio of the pixel point;

[0046] S22. Calculate the color entropy of the latest multispectral image based on the color ratios of all pixels.

[0047] In the present invention, the color rate of the pixel point is obtained by comparing the pixel values of the red, yellow and green bands with the maximum pixel value of each band and summing them up. It can comprehensively reflect the relative intensity of the pixel point in different color channels and provide a comprehensive indicator for further analysis of color characteristics.

[0048] In this embodiment of the present invention, S22 includes the following sub-steps:

[0049] S221, calculating a nonlinear adjustment value for each band based on the lower quartile and upper quartile of the pixel values of all pixels of the latest multispectral image in each band;

[0050] S222, determining an adjustment range of the latest multispectral image according to the nonlinear adjustment value of each band;

[0051] S223, determining the ratio of the number of pixels whose color ratios fall within the adjustment range to the total number of pixels in the latest multispectral image;

[0052] S224 , determining the color entropy of the latest multispectral image according to the ratio of the number of pixels whose color ratios fall within the adjustment range to the total number of pixels in the latest multispectral image.

[0053] In this paper, the use of upper and lower quartiles to calculate nonlinear adjustment values avoids interference from extreme values, making the data more robust and adaptable to complex image features. The adjustment range combines the adjustment values for each band with the maximum pixel value to define the effective range of color ratios and screen representative pixels. Color entropy is calculated based on the proportion of pixels whose color ratios fall within the adjustment range. This pixel distribution ratio is converted into a measure of color distribution uncertainty, providing effective features for the classification of pests, diseases, and dead trees.

[0054] In the embodiment of the present invention, in S221, the nonlinear adjustment value of the band The calculation formula is:

[0055] ;

[0056] Where, Indicates the quartiles of the pixel values of all pixels in the latest multispectral image in the band. Indicates the lower quartile of the pixel values of all pixels in the latest multispectral image in the band. Indicates the maximum pixel value of all pixels in the latest multispectral image in the band, Indicates the minimum pixel value of all pixels in the latest multispectral image in the band, Represents an exponent.

[0057] In the embodiment of the present invention, in S222, the expression of the adjustment range of the latest multispectral image is: ,in, Indicates the nonlinear adjustment value of the red band, Indicates the nonlinear adjustment value of the green band, Indicates the nonlinear adjustment value of the blue band, Indicates finding the minimum value, Indicates the maximum pixel value of all pixels in all bands in the latest multispectral image.

[0058] In the embodiment of the present invention, in S224, the color entropy of the latest multispectral image The calculation formula is:

[0059] ;

[0060] Where, represents the logarithmic function, Indicates the ratio of the number of pixels whose color ratio falls within the adjustment range to the total number of pixels in the latest multispectral image.

[0061] In this embodiment of the present invention, S3 includes the following sub-steps:

[0062] S31, taking the pixel points whose normalized red edge index in the latest multispectral image is less than the set threshold as the diseased and insect-infested tree area;

[0063] S32, based on the remaining areas except the pest-infested tree areas in the latest multispectral image, constructing a screening objective function using the color entropy generated by the latest multispectral image;

[0064] S33, determining the optimal target pixel point using the screening objective function;

[0065] S34. Pixels in the remaining area whose Manhattan distance to the optimal target pixel is less than a set threshold are regarded as dead tree areas.

[0066] In the present invention, the Normalized Red Edge Index is a remote sensing indicator used to assess the health and growth status of vegetation. The decrease in the NDRE value (Normalized Red Edge Index) is directly related to the reduction of chlorophyll caused by pests and diseases, and can be used as an effective screening indicator. Pixels with a value less than the set threshold are regarded as areas of pest-infested trees. Specific indicators can be used to quickly locate areas that may be affected by pests and diseases, providing a basis for further distinguishing dead trees. A screening objective function is constructed based on the remaining areas excluding the areas of pest-infested trees. This function comprehensively considers the relationship between color entropy and the mean of pixel values in different bands, and can highlight points in the remaining areas that have large differences in color characteristics from other pixels, providing an effective mathematical model for screening dead tree areas. The screening objective function is used to determine the optimal target pixel point, which can be regarded as a representative pixel point in the remaining area that is most relevant to the characteristics of dead trees. The pixels in the remaining area whose Manhattan distance to the best target pixel is less than the set threshold are regarded as the dead tree area. The range of dead trees is determined by measuring the color distance. Taking into account the color similarity of dead trees, the dead tree area can be identified more accurately, avoiding the uncertainty and randomness of single pixel judgment.

[0067] In the embodiment of the present invention, in S32, the expression of the screening objective function is:

[0068] ;

[0069] Where, represents the color entropy of the latest multispectral image, Indicates the remaining area The average pixel value of a pixel in all bands, represents the best target pixel point determined by screening the objective function, Represents the mean pixel value of all pixels in the remaining area in the red band, Represents the mean pixel value of all pixels in the remaining area in the green band, Represents the mean pixel value of all pixels in the remaining area in the blue band.

[0070] Those skilled in the art will appreciate that the embodiments described herein are intended to help readers understand the principles of the present invention, and it should be understood that the scope of protection of the present invention is not limited to such specific descriptions and embodiments. Those skilled in the art can make various other specific variations and combinations based on the technical teachings disclosed in the present invention without departing from the essence of the present invention, and such variations and combinations are still within the scope of protection of the present invention.

Claims

1. An intelligent classification method for pest-infested trees and dead trees based on multispectral images, characterized by: The following steps are involved: S1. Collect multispectral images of the area to be classified, perform geometric correction and radiation correction on the multispectral images, and then extract the pixel values of the pixels in each band in the latest multispectral images; S2, determining the nonlinear adjustment value of each band according to the pixel value of each pixel point in the latest multispectral image in each band, and generating color entropy for the latest multispectral image; S3, determining the pest and disease area of the latest multispectral image, and determining the dead tree area in the remaining area based on the color entropy of the latest multispectral image; The S2 includes the following sub-steps: S21. Adding the ratio of the pixel value of the pixel point in the latest multispectral image in the red band to the maximum red band pixel value, the ratio of the pixel value in the blue band to the maximum blue band pixel value, and the ratio of the pixel value in the green band to the maximum green band pixel value to obtain the color ratio of the pixel point; S22. Calculate the color entropy of the latest multispectral image based on the color ratios of all pixels; The S22 includes the following sub-steps: S221, calculating a nonlinear adjustment value for each band based on the lower quartile and upper quartile of the pixel values of all pixels of the latest multispectral image in each band; S222, determining an adjustment range of the latest multispectral image according to the nonlinear adjustment value of each band; S223, determining the ratio of the number of pixels whose color ratios fall within the adjustment range to the total number of pixels in the latest multispectral image; S224, determining the color entropy of the latest multispectral image based on the ratio of the number of pixels whose color ratios fall within the adjustment range to the total number of pixels in the latest multispectral image; In the step S221, the nonlinear adjustment value of the band The calculation formula is: ； Where, Indicates the quartiles of the pixel values of all pixels in the latest multispectral image in the band. Indicates the lower quartile of the pixel values of all pixels in the latest multispectral image in the band. Indicates the maximum pixel value of all pixels in the latest multispectral image in the band, Indicates the minimum pixel value of all pixels in the latest multispectral image in the band, represents the index; In S222, the expression of the adjustment range of the latest multispectral image is: ,in, Indicates the nonlinear adjustment value of the red band, Indicates the nonlinear adjustment value of the green band, Indicates the nonlinear adjustment value of the blue band, Indicates finding the minimum value, Indicates the maximum pixel value of all pixels in all bands in the latest multispectral image; In S224, the color entropy of the latest multispectral image The calculation formula is: ； Where, represents the logarithmic function, Indicates the ratio of the number of pixels whose color ratio falls within the adjustment range to the total number of pixels in the latest multispectral image; The S3 includes the following sub-steps: S31, taking the pixel points whose normalized red edge index in the latest multispectral image is less than the set threshold as the diseased and insect-infested tree area; S32, based on the remaining areas except the pest-infested tree areas in the latest multispectral image, constructing a screening objective function using the color entropy generated by the latest multispectral image; S33, determining the optimal target pixel point using the screening objective function; S34, taking pixels in the remaining area whose Manhattan distance to the optimal target pixel is less than a set threshold as dead tree areas; In the S32, the expression of the screening objective function is: ； Where, represents the color entropy of the latest multispectral image, Indicates the remaining area The average pixel value of a pixel in all bands, represents the best target pixel point determined by screening the objective function, Represents the mean pixel value of all pixels in the remaining area in the red band, Represents the mean pixel value of all pixels in the remaining area in the green band, Represents the mean pixel value of all pixels in the remaining area in the blue band.

Images