Remote sensing image sugarcane planting field identification method based on deep convolutional neural network

Abstract

The invention relates to the technical field of remote sensing images, in particular to a remote sensing image sugarcane planting field identification method based on a deep convolutional neural network, which comprises the following steps of: acquiring satellite remote sensing image data of a plurality of sugarcane planting fields from a network database, analyzing the acquired satellite remote sensing image data of the sugarcane planting fields, and acquiring a sugarcane planting field identification result; the method comprises the following steps: acquiring satellite remote sensing image data of a plurality of sugarcane planting fields from a network database, analyzing the acquired satellite remote sensing image data of the sugarcane planting fields, comparing image data needing to be identified with image data selected in the network database, and identifying the similarity of two groups of remote sensing image data; satellite remote sensing image data with the highest similarity with image data needing to be recognized is selected for processing, errors of remote sensing image data due to differences of sun position and angle conditions, mist, haze and other atmospheric conditions and the like are eliminated as much as possible, and recognition is more accurate.

Description

Technical Field

[0001] The present invention belongs to the technical field of remote sensing image recognition, and in particular relates to a remote sensing image sugarcane planting site recognition method based on a deep convolutional neural network. Background Art

[0002] Remote sensing image recognition refers to the use of computers to analyze the spectral and spatial information of various types of objects in remote sensing images, select features, and use certain means to divide the feature space into non-overlapping subspaces, and then classify each pixel in the image into the corresponding object category in each subspace. Remote sensing image recognition can realize the perception and diagnosis of agricultural data, and ultimately achieve precision planting and intelligent management.

[0003] In order to understand the scale of sugarcane planting in real time during sugarcane planting, the satellite remote sensing images taken can be identified through remote sensing image recognition technology to further identify the scale of sugarcane planting. However, the existing remote sensing image recognition technology is easily affected by differences in the position and angle of the sun, mist, and other atmospheric conditions, which may lead to errors in recognition. Summary of the invention

[0004] The purpose of the present invention is to solve the shortcomings of the prior art and propose a method for identifying sugarcane planting areas in remote sensing images based on a deep convolutional neural network.

[0005] In order to achieve the above object, the present invention adopts the following technical solutions: A method for identifying sugarcane planting areas in remote sensing images based on deep convolutional neural networks is designed, which includes the following steps: S1. Obtain satellite remote sensing image data of multiple sugarcane planting areas from a network database, analyze the acquired satellite remote sensing image data of the sugarcane planting areas, compare the image data to be identified with the image data selected from the network database, identify the similarity of the two sets of remote sensing image data, and select the satellite remote sensing image data with the highest similarity to the image data to be identified for processing; S2, the selected satellite remote sensing image data is processed and divided into training set and test set to build a deep convolutional neural network model; S3, input the training set into the deep convolutional neural network model for training, input the test set into the trained deep convolutional neural network model for testing, obtain the performance and accuracy evaluation of the model, and then obtain a deep convolutional neural network model that passes the test; S4. Input the image data to be identified and detected into the deep convolutional neural network model, and the deep convolutional neural network model outputs the identification result of the sugarcane planting area after calculation.

[0006] Preferably, when selecting the image data of the multiple sugarcane planting sites acquired in S1, images of sugarcane at different planting stages are selected.

[0007] Preferably, the image data to be identified are compared with the image data selected from the network database in terms of their spatial resolution, spectral resolution, radiation resolution and time resolution, so that the shooting heights and shooting months of the two sets of data are as close as possible.

[0008] Preferably, the deep convolutional neural network model adopts data preprocessing techniques such as noise removal, input data dimensionality reduction and irrelevant data deletion in order to accelerate the convergence speed of the training algorithm.

[0009] Preferably, the satellite remote sensing image data of multiple sugarcane planting sites obtained in the network database removes information data irrelevant to the sugarcane fields during processing.

[0010] Preferably, the satellite remote sensing image data of the sugarcane planting area obtained from the network database is remote sensing image data with less atmospheric influence as much as possible.

[0011] The beneficial effects of the present invention are as follows: the present invention obtains satellite remote sensing image data of multiple sugarcane planting areas from a network database, analyzes the acquired satellite remote sensing image data of the multiple sugarcane planting areas, compares the image data to be identified with the image data selected from the network database, identifies the similarity of the two groups of remote sensing image data, selects the satellite remote sensing image data with the highest similarity to the image data to be identified for processing, and thereby eliminates errors in the remote sensing image data caused by differences in the sun's position and angle conditions, mist, haze and other atmospheric conditions as much as possible, thereby making the identification more accurate. DETAILED DESCRIPTION

[0012] The technical solutions in the embodiments of the present invention will be clearly and completely described below in combination with the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.

[0013] A method for identifying sugarcane planting areas in remote sensing images based on a deep convolutional neural network comprises the following steps: S1. Obtain satellite remote sensing image data of multiple sugarcane planting areas from a network database, analyze the acquired satellite remote sensing image data of the sugarcane planting areas, compare the image data to be identified with the image data selected from the network database, identify the similarity of the two sets of remote sensing image data, and select the satellite remote sensing image data with the highest similarity to the image data to be identified for processing; S2, the selected satellite remote sensing image data is processed and divided into training set and test set to build a deep convolutional neural network model; S3, input the training set into the deep convolutional neural network model for training, input the test set into the trained deep convolutional neural network model for testing, obtain the performance and accuracy evaluation of the model, and then obtain a deep convolutional neural network model that passes the test; S4. Input the image data to be identified and detected into the deep convolutional neural network model, and the deep convolutional neural network model outputs the identification result of the sugarcane planting area after calculation.

[0014] When selecting the image data of multiple sugarcane planting sites obtained in S1, images of sugarcane at different planting stages will be selected. There are large differences between the images of sugarcane at different planting stages. There will be deviations when the planting time is not added when planting sugarcane. This results in the need to integrate sugarcane image data at multiple stages for identification when identifying sugarcane fields in a large area.

[0015] When comparing the image data to be identified with the image data selected from the network database, the spatial resolution, spectral resolution, radiation resolution and time resolution are compared and identified, so that the shooting heights and shooting months of the two sets of data are as close as possible, so that the data of the two images can be closer. This method can effectively increase the accuracy of image recognition.

[0016] In order to accelerate the convergence speed of the training algorithm, the deep convolutional neural network model adopts data preprocessing techniques such as noise removal, input data dimensionality reduction and irrelevant data deletion. The satellite remote sensing image data of multiple sugarcane plantations are obtained from the network database. Information data irrelevant to the sugarcane fields is removed during processing. The parts of the image that are not related to the sugarcane fields are deleted through preprocessing technology, so that the algorithm can calculate faster.

[0017] The satellite remote sensing image data of sugarcane planting areas obtained from the network database try to use remote sensing image data with less atmospheric influence. Satellite remote sensing images will be affected by the atmosphere during shooting, which will affect their accuracy.

[0018] Specifically: satellite remote sensing image data of multiple sugarcane planting areas are obtained from a network database, the obtained satellite remote sensing image data of the sugarcane planting areas are analyzed, the image data to be identified is compared with the image data selected in the network database, the similarity of the two groups of remote sensing image data is identified, and the satellite remote sensing image data with the highest similarity to the image data to be identified is selected for processing. The selected satellite remote sensing image data are divided into a training set and a test set after processing, and a deep convolutional neural network model is constructed. The training set is input into the deep convolutional neural network model for training, and the test set is input into the trained deep convolutional neural network model for testing to obtain the performance and accuracy evaluation of the model, and then a deep convolutional neural network model that passes the test is obtained, and the picture data to be identified and detected is input into the deep convolutional neural network model. After the deep convolutional neural network model is operated, the recognition result of the sugarcane planting area is output.

[0019] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A method for identifying sugarcane planting areas in remote sensing images based on deep convolutional neural networks, characterized in that: The following steps are involved: S1. Obtain satellite remote sensing image data of multiple sugarcane planting areas from a network database, analyze the acquired satellite remote sensing image data of the sugarcane planting areas, compare the image data to be identified with the image data selected from the network database, identify the similarity of the two sets of remote sensing image data, and select the satellite remote sensing image data with the highest similarity to the image data to be identified for processing; S2, the selected satellite remote sensing image data is processed and divided into training set and test set to build a deep convolutional neural network model; S3, input the training set into the deep convolutional neural network model for training, input the test set into the trained deep convolutional neural network model for testing, obtain the performance and accuracy evaluation of the model, and then obtain a deep convolutional neural network model that passes the test; S4. Input the image data to be identified and detected into the deep convolutional neural network model, and the deep convolutional neural network model outputs the identification result of the sugarcane planting area after calculation.

2. The method for identifying sugarcane planting areas in remote sensing images based on a deep convolutional neural network according to claim 1, characterized in that: When selecting the image data of the multiple sugarcane planting sites acquired in S1, images of sugarcane at different planting stages will be selected.

3. The method for identifying sugarcane planting areas in remote sensing images based on deep convolutional neural networks according to claim 1, characterized in that: When the image data to be identified is compared with the image data selected from the network database, the spatial resolution, spectral resolution, radiation resolution and time resolution are compared and identified to make the shooting heights and shooting months of the two sets of data as close as possible.

4. The method for identifying sugarcane planting areas in remote sensing images based on deep convolutional neural networks according to claim 1, characterized in that: The deep convolutional neural network model adopts data preprocessing techniques such as noise removal, input data dimensionality reduction and irrelevant data deletion in order to accelerate the convergence speed of the training algorithm.

5. The method for identifying sugarcane planting areas in remote sensing images based on deep convolutional neural networks according to claim 1, characterized in that: The satellite remote sensing image data of multiple sugarcane planting areas are obtained from the network database, and information data irrelevant to the sugarcane planting areas is removed during processing.

6. The method for identifying sugarcane planting areas in remote sensing images based on deep convolutional neural networks according to claim 1, characterized in that: The satellite remote sensing image data of the sugarcane planting area obtained in the network database is selected as much as possible from remote sensing image data with less atmospheric influence.