An invasive species recognition method, system and storage medium based on multispectral image processing

Abstract

The present invention provides a method, system and storage medium for identifying invasive species based on multispectral image processing, which relates to the technical field of image processing. The method includes the following steps: acquiring multispectral image data of an area to be identified and preprocessing the multispectral image data; extracting the preprocessed multispectral image data to obtain spectral data corresponding to each pixel point in the multispectral image data; processing the spectral data to obtain a corresponding spectral data processing result, and at the same time processing the multispectral image data to obtain a corresponding image data processing result; fusing the spectral data processing result and the image data processing result, and determining whether there are invasive species according to the fusion result. The present invention can effectively improve the recognition accuracy.

Description

Technical Field

[0001] The present invention relates to the technical field of image processing, and more specifically, to an invasive species recognition method, system and storage medium based on multi-spectral image processing. Background Art

[0002] At present, an alien invasive species refers to a species, subspecies or lower taxonomic group that appears outside its natural distribution range and distribution location, including any part, gamete or propagule of these species that can survive and reproduce. The invasive alien species may damage the naturalness and integrity of the landscape, destroy the ecosystem, endanger the diversity of animals and plants, and affect the genetic diversity.

[0003] However, since invasive species, such as most alien plant varieties, usually live in the wild, the wild environment is relatively complex and the invasive plant varieties grow mixed with other plants, and the colors, structures and other characteristics of the alien plant varieties are relatively similar to those of other plants. Existing invasive species recognition methods may not be able to identify invasive species from images, resulting in problems of low recognition accuracy and low recognition efficiency.

[0004] Therefore, how to provide an invasive species recognition method that can solve the above problems is an urgent problem for those skilled in the art. Summary of the Invention

[0005] In view of this, the present invention provides an invasive species recognition method, system and storage medium based on multi-spectral image processing. By processing multi-spectral image data and determining whether there are invasive species according to the processing results of the multi-spectral image data, the recognition accuracy can be effectively improved.

[0006] To achieve the above object, the present invention adopts the following technical solutions:

[0007] An invasive species recognition method based on multi-spectral image processing includes the following steps:

[0008] Obtain multi-spectral image data of the area to be recognized, and preprocess the multi-spectral image data;

[0009] Extract the preprocessed multi-spectral image data to obtain spectral data corresponding to each pixel point in the multi-spectral image data;

[0010] Process the spectral data to obtain corresponding spectral data processing results, and at the same time process the multi-spectral image data to obtain corresponding image data processing results;

[0011] Fuse the spectral data processing results and the image data processing results, and determine whether there are invasive species according to the fusion results.

[0012] Preferably, the specific process of obtaining the corresponding spectral data processing result includes:

[0013] Performing clustering processing on the spectral data to obtain a corresponding spectral clustering result;

[0014] Constructing a pattern recognition model, and inputting the spectral clustering result into the pattern recognition model for processing to obtain a corresponding spectral data processing result.

[0015] Preferably, the specific processing process of obtaining the corresponding image data processing result includes:

[0016] Successively performing region of interest extraction and feature extraction on the preprocessed multi-spectral image data;

[0017] Using the feature extraction result to construct a feature vector;

[0018] Establishing a recognition model, inputting the feature vector into the recognition model for recognition to obtain a corresponding image data processing result.

[0019] Preferably, the specific processing process of determining whether there is an invasive species according to the fusion result includes:

[0020] Matching the fusion result with a preset species image database, and determining the invasive species according to the matching result.

[0021] Preferably, the specific processing process of obtaining the fusion result includes:

[0022] Performing interpolation and weighted fusion processing on the spectral data processing result and the image data processing result.

[0023] The present invention also provides an invasive species recognition system based on multi-spectral image processing, including:

[0024] An acquisition module, configured to acquire multi-spectral image data of a region to be recognized, and preprocess the multi-spectral image data;

[0025] An extraction module, configured to extract the preprocessed multi-spectral image data to obtain spectral data corresponding to each pixel point in the multi-spectral image data;

[0026] A processing module, configured to process the spectral data to obtain a corresponding spectral data processing result, and simultaneously process the multi-spectral image data to obtain a corresponding image data processing result;

[0027] A judgment module, configured to fuse the spectral data processing result and the image data processing result, and determine whether there is an invasive species according to the fusion result.

[0028] The present invention also provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the invasive species recognition method described in any one of the above is implemented.

[0029] As can be seen from the above technical solutions, compared with the prior art, the present invention discloses an invasive species recognition method, system and storage medium based on multi-spectral image processing. By collecting multi-spectral image data of the area to be recognized, preprocessing the multi-spectral image data, obtaining corresponding spectral data according to the preprocessed multi-spectral image data, and respectively processing the spectral data and the multi-spectral image data, then fusing the spectral data processing result and the multi-spectral image processing result and matching with a preset species image database, and determining the invasive species according to the matching result, the accuracy and efficiency of invasive species recognition can be improved, the false alarm rate and the missed alarm rate can be reduced, it is helpful to detect potential invasive species more quickly and accurately, improve the degree of automation, reduce the cost of recognition and monitoring, and bring more economic benefits to ecological protection and management. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required to be used in the description of the embodiments or the prior art. Obviously, the drawings in the following description are only the embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained according to the provided drawings without creative efforts.

[0031] Figure 1 It is the overall flowchart of an invasive species recognition method based on multi-spectral image processing provided by the present invention;

[0032] Figure 2 It is the structural principle block diagram of an invasive species recognition system based on multi-spectral image processing provided by the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0033] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0034] See Figure 1 As shown, the embodiments of the present invention disclose an invasive species recognition method based on multi-spectral image processing, including the following steps:

[0035] Obtain the multispectral image data of the area to be recognized, and preprocess the multispectral image data, where the multispectral image data can be achieved by a drone equipped with a high-definition remote sensing image acquisition device, and the preprocessing process may include filtering the image;

[0036] Extract the preprocessed multispectral image data to obtain the spectral data corresponding to each pixel point in the multispectral image data;

[0037] Process the spectral data to obtain the corresponding spectral data processing result, and at the same time process the multispectral image data to obtain the corresponding image data processing result;

[0038] Fuse the spectral data processing result and the image data processing result, and determine whether there are invasive species according to the fusion result.

[0039] In a specific embodiment, the specific process of obtaining the corresponding spectral data processing result includes:

[0040] Perform clustering processing on the spectral data to obtain the corresponding spectral clustering result;

[0041] Construct a pattern recognition model, and input the spectral clustering result into the pattern recognition model for processing to obtain the corresponding spectral data processing result.

[0042] Specifically, the pattern recognition model can be a support vector machine, and the clustering processing can be implemented by the k-means clustering method. Through the clustering processing, relatively approximate spectral data can be classified. Spectral data has many advantages in the field of remote sensing, and can provide various types of information of invasive species while having a high target discrimination degree to improve the recognition accuracy.

[0043] In a specific embodiment, the specific processing process of obtaining the corresponding image data processing result includes:

[0044] Successively perform region of interest extraction and feature extraction on the preprocessed multispectral image data;

[0045] Construct a feature vector using the feature extraction result;

[0046] Establish a recognition model, input the feature vector into the recognition model for recognition, and obtain the corresponding image data processing result.

[0047] Specifically, the recognition model can be implemented by a neural network. The feature vector is dimensionally reduced using an autoencoder network as the network backbone structure, then the dimensionally reduced feature vector is processed by a pyramid convolution module, and finally a decoder structure composed of a transposed convolution and a convolution module is applied for decoding to output the image processing result.

[0048] During the process of training the recognition model, first, obtain historical images of invasive species and historical images of native plants similar to invasive species. After annotating the historical images of invasive species and native plants, mix and perform sample balancing processing on them to form a training dataset. Then, divide the training dataset into a training set and a test set. During the process of training the model using the training set, compare the recognition results with the annotations to obtain the final recognition error. If the recognition error meets the preset threshold requirements, stop training and use the test set to test the recognition model. If the recognition error does not meet the preset threshold requirements, iteratively correct the misrecognized results and backpropagate them to the recognition model to update the network parameters of the recognition model, and then use the test set to test the recognition model, which can improve the subsequent model recognition accuracy.

[0049] In a specific embodiment, the specific process of determining whether there are invasive species according to the fusion result includes:

[0050] Match the fusion result with a preset species image database, and determine the invasive species according to the matching result.

[0051] In a specific embodiment, the specific process of obtaining the fusion result includes:

[0052] Perform interpolation and weighted fusion processing on the spectral data processing result and the image data processing result.

[0053] See Figure 2 As shown, the embodiment of the present invention also provides a recognition system using the invasive species recognition method based on multispectral image processing according to any one of the above embodiments, including:

[0054] An acquisition module, configured to acquire multispectral image data of the area to be recognized and preprocess the multispectral image data;

[0055] An extraction module, configured to extract the preprocessed multispectral image data to obtain the spectral data corresponding to each pixel point in the multispectral image data;

[0056] A processing module, configured to process the spectral data to obtain the corresponding spectral data processing result, and at the same time process the multispectral image data to obtain the corresponding image data processing result;

[0057] A judgment module, configured to fuse the spectral data processing result and the image data processing result, and determine whether there are invasive species according to the fusion result.

[0058] The embodiment of the present invention also provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it implements the invasive species recognition method according to any one of the above embodiments.

[0059] The various embodiments in this specification are described in a progressive manner. Each embodiment focuses on the differences from other embodiments. For the same or similar parts among the embodiments, reference can be made to each other. For the devices disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple. For the relevant parts, reference can be made to the description in the method section.

[0060] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present invention. Various modifications to these embodiments will be obvious to those skilled in the art. The general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to these embodiments shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. An invasive species recognition method based on multispectral image processing, characterized in that, It includes the following steps: Obtain the multi-spectral image data of the area to be recognized, and preprocess the multi-spectral image data; Extract the preprocessed multi-spectral image data to obtain the spectral data corresponding to each pixel point in the multi-spectral image data; Process the spectral data to obtain the corresponding spectral data processing result, and at the same time process the multi-spectral image data to obtain the corresponding image data processing result; Fuse the spectral data processing result and the image data processing result, and determine whether there is an invasive species according to the fusion result; The specific process of obtaining the corresponding spectral data processing result includes: Perform clustering processing on the spectral data to obtain the corresponding spectral clustering result; Construct a pattern recognition model, and input the spectral clustering result into the pattern recognition model for processing to obtain the corresponding spectral data processing result; The specific processing process of determining whether there is an invasive species according to the fusion result includes: Match the fusion result with the preset species image database, and determine the invasive species according to the matching result; The recognition model is implemented by a neural network. The feature vector is dimension-reduced with an autoencoder network as the network backbone structure, then the dimension-reduced feature vector is processed by a pyramid convolution module, and finally a decoder structure composed of a transposed convolution and a convolution module is applied for decoding to output the image processing result; During the training process of the recognition model, first obtain the historical images of invasive species and the historical images of native plants similar to the invasive species, and after mixing and sample balancing processing the historical images of invasive species and native plants after annotation, form a training data set, and then divide the training data set into a training set and a test set. During the process of training the model with the training set, compare the recognition result with the annotation to obtain the final recognition error. If the recognition error meets the preset threshold requirement, stop training and test the recognition model with the test set. If the recognition error does not meet the preset threshold requirement, iterate and correct the recognition error result and then backpropagate it to the recognition model to update the network parameters of the recognition model and then test the recognition model with the test set.

2. The invasive species identification method based on multispectral image processing according to claim 1, characterized in that The specific processing process of obtaining the corresponding image data processing result includes: Successively perform region of interest extraction and feature extraction on the preprocessed multi-spectral image data; Construct a feature vector using the feature extraction result; Establish a recognition model, input the feature vector into the recognition model for recognition, and obtain the corresponding image data processing result.

3. The invasive species recognition method based on multispectral image processing according to claim 1, characterized in that, The specific processing process of obtaining the fusion result includes: Perform interpolation and weighted fusion processing on the spectral data processing result and the image data processing result.

4. An identification system using the invasive species identification method based on multispectral image processing according to any one of claims 1-3, characterized in that, It includes: An acquisition module, used to obtain the multi-spectral image data of the area to be recognized, and preprocess the multi-spectral image data; An extraction module, used to extract the preprocessed multi-spectral image data to obtain the spectral data corresponding to each pixel point in the multi-spectral image data; A processing module, configured to process the spectral data to obtain a corresponding spectral data processing result, and at the same time process the multispectral image data to obtain a corresponding image data processing result; A judgment module, configured to fuse the spectral data processing result and the image data processing result, and determine whether there is an invasive species according to the fusion result.

5. A computer-readable storage medium, characterized in that, A computer program is stored on the computer-readable storage medium, and when the computer program is executed by a processor, the invasive species recognition method according to any one of claims 1 to 3 is implemented.

Images