Soil deep scarification cultivation method based on artificial intelligence system
By using an artificial intelligence-based deep tillage method that incorporates visible light image processing and deep learning networks, the problem of precise control in traditional deep tillage has been solved, thereby improving soil structure and crop growth.
Patent Information
- Application Number
- CN202410973398.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-19
- Publication Date
- 2024-11-08
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
Traditional deep tillage methods rely on labor experience, making it difficult to accurately control the depth and frequency of tillage, which leads to soil structure damage, soil erosion, and decreased fertility.
An artificial intelligence-based deep tillage method is adopted. Visible light images of soil samples are acquired, brightness components are calculated for image enhancement, features are extracted using convolutional and pooling layers, and a soil type prediction model is trained using a deep learning network to determine the appropriate deep tillage method.
It enables more accurate soil type identification, improves soil structure, increases soil aeration and water retention, and promotes crop growth.
Smart Images

Figure CN118918373A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of deep soil tillage, and in particular to a deep soil tillage method based on an artificial intelligence system. Background Art
[0002] Agriculture is the foundation of human civilization, and the development of farming techniques is directly linked to ecological sustainability. Traditional tillage methods are a key measure for improving soil aeration and permeability, texture, and fertility. However, traditional deep tillage relies on experience, making it difficult to precisely control tillage depth and frequency. This can lead to soil structural damage, soil erosion, and decreased fertility, and therefore presents certain limitations. Summary of the Invention
[0003] In order to solve the above problems, the purpose of the present invention is to provide a soil deep tillage method based on an artificial intelligence system.
[0004] A soil deep tillage method based on an artificial intelligence system, comprising:
[0005] Step 1: Soil samples are collected from different types of cultivated land and visible light images of the soil samples are acquired;
[0006] Step 2: Calculate the brightness component of the visible light image, and perform detail enhancement on the visible light image based on the brightness component to obtain a detail-enhanced visible light image;
[0007] Step 3: Use convolutional layers and pooling layers to extract the overall and local features of the visible light image;
[0008] Step 4: Combine the overall features and local features to obtain the global features, and input the global features into the deep learning network for training to obtain the cultivated land soil type prediction model;
[0009] Step 5: using the farmland soil type prediction model to determine the soil type of the target farmland;
[0010] Step 6: Determine the soil deep loosening tillage method based on the soil type in the target cultivated land.
[0011] Preferably, the step 2: calculating the brightness component of the visible light image, and performing detail enhancement of the visible light image based on the brightness component to obtain a detail-enhanced visible light image, includes:
[0012] Step 2.1: Determine the brightness component of the visible light image based on the red, green, and blue color channel values of the visible light image; the brightness component calculation formula is:
[0013] Y=0.299R+0.587G+0.114B
[0014] Among them, Y represents the brightness component, R represents the red color channel value, G represents the green color channel value, and B represents the blue color channel value;
[0015] Step 2.2: performing logarithmic transformation on the luminance component to obtain a luminance logarithmic output;
[0016] Step 2.3: Smoothing the visible light image based on the brightness logarithmic output to obtain a smoothed and enhanced image;
[0017] Step 2.4: Introduce contrast parameters to perform global detail enhancement on the smoothed enhanced image to obtain a detail-enhanced visible light image.
[0018] Preferably, the step 2.2: performing logarithmic transformation on the luminance component to obtain a luminance logarithmic output comprises:
[0019] Using the formula:
[0020]
[0021] Perform logarithmic transformation on the brightness component to obtain the brightness logarithmic output; where L(x,y) represents the brightness component at the (x,y)th point, M is the maximum value of the pixel range of the image, and L g (x,y) represents the logarithmic output of the brightness at the (x,y)th point.
[0022] Preferably, the step 2.3 of smoothing the visible light image based on the brightness logarithmic output to obtain a smoothed and enhanced image includes:
[0023] An image smoothing enhancement formula is constructed based on a guided filter, and the image smoothing enhancement formula is used to smooth the visible light image to obtain a smoothed enhanced image; wherein the image smoothing enhancement formula is:
[0024]
[0025] Among them, L y (x,y) represents the image after smooth enhancement, H g (x,y) represents the guided filter, (ξ x ,ξ y ) represents the pixel coordinates of the domain, ω(x,y) represents the square area with a side length of r centered at the pixel point (x,y), |ω| represents the number of pixels within ω(x,y), a(ξ x ,ξ y ) represents the first adjustment coefficient
[0026] b(ξ x ,ξ y )=μ(ξx ,ξ y )-a(ξ x ,ξ y )μ(ξ x ,ξ y ), μ2(ξ x ,ξ y )express
[0027] The mean of the square of the logarithmic output of brightness within ω(x,y), μ(ξ x ,ξ y ) represents the mean value of the logarithmic output of brightness within ω(x,y), σ 2 (ξ x ,ξ y ) represents the variance of the logarithmic output of brightness within ω(x,y), and ε represents the constant coefficient.
[0028] Preferably, in step 2.4, the formula is used:
[0029]
[0030] Perform global detail enhancement on the smoothed enhanced image to obtain a detail-enhanced visible light image; where α(x, y) represents the contrast parameter, η represents the parameter for image enhancement control, and L gmax represents the maximum value of the brightness logarithm output, λ represents the empirical parameter, represents the mean of the global logarithmic transformation, δ is a constant, and N represents the total number of pixels in the image.
[0031] Preferably, in step 3, the visible light image is sequentially input into the convolution layer and the pooling layer to obtain the overall feature; wherein the overall feature extraction formula is:
[0032] c i =f(ω·H+b)
[0033]
[0034] Where f is the activation function, ω is the convolution kernel, h is the convolution kernel size, b is the bias, c is the i is the feature vector extracted by the i-th convolutional layer, It represents the value of the features extracted by three different convolution kernels after the maximum pooling layer, and C represents the overall features extracted.
[0035] Preferably, in step 3, the local feature calculation formula is:
[0036]
[0037] Among them, F represents the visible light image, F1 represents the first feature map, F2 represents the second feature map, F3 represents the third feature map, F4 represents the fourth feature map, F″ represents the local feature, Conv1, Conv3, Conv4 and Conv6 represent 1×1×1 convolution, Conv2 and Conv5 represent 3×3×3 convolution, Cat represents the splicing operation, and Maxpoo represents the maximum pooling.
[0038] Preferably, in step 4, a loss function is used to optimize the deep learning network and train it to obtain a cultivated land soil type prediction model; wherein the loss function is:
[0039]
[0040] in, represents the true category of soil type, m represents the total number of soil type categories, λ represents the regularization parameter, y k Represents the prediction result output by the deep learning network, and θ represents the network parameters.
[0041] The present invention also provides an electronic device, comprising a bus, a transceiver, a memory, a processor, and a computer program stored in the memory and runnable on the processor, wherein the transceiver, the memory, and the processor are connected via the bus, and wherein the computer program, when executed by the processor, implements the steps of the above-mentioned soil deep loosening and tillage method based on an artificial intelligence system.
[0042] The present invention also provides a computer-readable storage medium having a computer program stored thereon, characterized in that when the computer program is executed by a processor, the steps in the above-mentioned soil deep loosening and tillage method based on an artificial intelligence system are implemented.
[0043] The beneficial effect of the soil deep loosening and tillage method based on the artificial intelligence system provided by the present invention is that: compared with the existing technology, the present invention can more accurately determine the soil type in the target cultivated land by training the deep learning network using visible light images of soil samples, thereby determining the appropriate deep loosening and tillage method according to the soil type, which helps to improve soil structure, increase soil permeability and water retention, and promote crop growth.
[0044] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, preferred embodiments are given below and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS
[0045] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.
[0046] Figure 1 A flow chart of a soil subsoiling method based on an artificial intelligence system provided by an embodiment of the present invention is shown;
[0047] Figure 2 The figure shows the principle diagram of local feature extraction provided by an embodiment of the present invention. DETAILED DESCRIPTION
[0048] In the description of the present invention, it should be understood that the terms "center", "longitudinal", "lateral", "length", "width", "thickness", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", "clockwise", "counterclockwise" and the like to indicate orientations or positional relationships based on the orientations or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be understood as limiting the present invention.
[0049] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of the technical features being referred to. Thus, a feature identified as "first" or "second" may explicitly or implicitly include one or more of the features. In the description of the present invention, "plurality" means two or more, unless otherwise specifically defined.
[0050] In the present invention, unless otherwise expressly specified or limited, the terms "mounted," "connected," "connect," "fixed," etc. should be understood broadly. For example, they may refer to fixed, detachable, or integral connections; mechanical or electrical connections; direct or indirect connections through an intermediary; or internal communication between two components. Those skilled in the art will understand the specific meanings of the above terms in the present invention based on specific circumstances.
[0051] See also Figure 1 , a soil deep tillage method based on an artificial intelligence system, comprising:
[0052] Step 1: Soil samples are collected from different types of cultivated land and visible light images of the soil samples are acquired;
[0053] Step 2: Calculate the brightness component of the visible light image, and perform detail enhancement on the visible light image based on the brightness component to obtain a detail-enhanced visible light image;
[0054] Furthermore, the step 2 includes:
[0055] Step 2.1: Determine the brightness component of the visible light image based on the red, green, and blue color channel values of the visible light image; the brightness component calculation formula is:
[0056] Y=0.299R+0.587G+0.114B
[0057] Among them, Y represents the brightness component, R represents the red color channel value, G represents the green color channel value, and B represents the blue color channel value;
[0058] Step 2.2: performing logarithmic transformation on the luminance component to obtain a luminance logarithmic output;
[0059] In step 2, the present invention may adopt the formula:
[0060]
[0061] Perform logarithmic transformation on the brightness component to obtain the brightness logarithmic output; where L(x,y) represents the brightness component at the (x,y)th point, M is the maximum value of the pixel range of the image, and L g (x,y) represents the logarithmic output of the brightness at the (x,y)th point.
[0062] Step 2.3: Smoothing the visible light image based on the brightness logarithmic output to obtain a smoothed and enhanced image;
[0063] In step 2.3, an image smoothing enhancement formula is constructed based on the guided filter, and the image smoothing enhancement formula is used to smooth the visible light image to obtain a smoothed enhanced image; wherein the image smoothing enhancement formula is:
[0064]
[0065] Among them, L y (x,y) represents the image after smooth enhancement, H g (x,y) represents the guided filter, (ξ x ,ξ y ) represents the pixel coordinates of the domain, ω(x,y) represents the square area with a side length of r centered at the pixel point (x,y), |ω represents the number of pixels within ω(x,y), a(ξ x ,ξ y ) represents the first adjustment coefficient
[0066] b(ξ x ,ξ y )=μ(ξ x ,ξ y )-a(ξ x ,ξ y )μ(ξ x ,ξ y ), μ2(ξ x ,ξ y )express
[0067] The mean of the square of the logarithmic output of brightness within ω(x,y), μ(ξ x ,ξ y ) represents the mean value of the logarithmic output of brightness within ω(x,y), σ 2 (ξ x ,ξ y ) represents the variance of the logarithmic output of brightness within ω(x,y), and ε represents the constant coefficient.
[0068] By performing a logarithmic transformation on the luminance component Y, the brightness of the image is improved, but the image details are not enhanced. By introducing a guided filter to replace the Gaussian filter in the traditional Retinex, the present invention can better preserve the details of the image edges, making the enhanced image more natural and clear, while also better distinguishing between noise and edges in the image.
[0069] Step 2.4: Introduce contrast parameters to perform global detail enhancement on the smoothed enhanced image to obtain a detail-enhanced visible light image.
[0070] After guided filtering, image halo artifacts are significantly reduced. However, the image contrast is low, and the output enhanced image is visually unsatisfactory. Due to the characteristics of the edge filter, it assigns very low weights to neighboring pixels with a large brightness difference from the central pixel, resulting in a relatively flat image. Therefore, a contrast enhancement factor is introduced to enhance image detail.
[0071] In step 2.4, the present invention may adopt the formula:
[0072]
[0073] Perform global detail enhancement on the smoothed enhanced image to obtain a detail-enhanced visible light image; where α(x, y) represents the contrast parameter, η represents the parameter for image enhancement control, and L gmax represents the maximum value of the brightness logarithm output, λ represents the empirical parameter, represents the mean of the global logarithmic transformation, δ is a constant, and N represents the total number of pixels in the image.
[0074] Step 3: Use convolutional layers and pooling layers to extract the overall and local features of the visible light image;
[0075] In step 3, the visible light image is sequentially input into the convolution layer and the pooling layer to obtain the overall feature; wherein, the overall feature extraction formula is:
[0076] c i =f(ω·H+b)
[0077]
[0078] Where f is the activation function, ω is the convolution kernel, h is the convolution kernel size, b is the bias, c is the i is the feature vector extracted by the i-th convolutional layer, It represents the value of the features extracted by three different convolution kernels after the maximum pooling layer, and C represents the overall features extracted.
[0079] See also Figure 2 , in step 3, the local feature calculation formula is:
[0080]
[0081] Wherein, F represents a visible light image, F1 represents the first feature map, F2 represents the second feature map, F3 represents the third feature map, F4 represents the fourth feature map, F″ represents a local feature, Conv1, Conv3, Conv4 and Conv6 represent 1×1×1 convolutions, Conv2 and Conv5 represent 3×3×3 convolutions, Cat represents a splicing operation, and Maxpoo represents maximum pooling. The present invention can capture local features in an image by pooling the visible light image at different scales.
[0082] Step 4: Combine the overall features and local features to obtain the global features, and input the global features into the deep learning network for training to obtain the cultivated land soil type prediction model;
[0083] In step 4, the deep learning network is optimized and trained using the loss function to obtain the cultivated land soil type prediction model; wherein the loss function is:
[0084]
[0085] in, represents the true category of soil type, m represents the total number of soil type categories, λ represents the regularization parameter, y k Represents the prediction result output by the deep learning network, and θ represents the network parameters.
[0086] By combining global and local features, this method can better capture both the global information and local details of an image, helping to improve the model's performance in complex scenarios. By comprehensively considering global features, the model can more accurately understand the relationships between pixels, thereby improving the prediction of arable land soil types.
[0087] Step 5: using the farmland soil type prediction model to determine the soil type of the target farmland;
[0088] Step 6: Determine the soil deep loosening tillage method based on the soil type in the target cultivated land.
[0089] By using visible light images of soil samples to train a deep learning network, the present invention can more accurately determine the soil type in the target cultivated land, thereby determining the appropriate deep loosening tillage method based on the soil type, which helps to improve soil structure, increase soil aeration and water retention, and promote crop growth.
[0090] The present invention also provides an electronic device, comprising a bus, a transceiver, a memory, a processor, and a computer program stored in the memory and runnable on the processor, wherein the transceiver, the memory, and the processor are connected via the bus, and is characterized in that when the computer program is executed by the processor, the steps of the above-mentioned soil deep loosening and tillage method based on an artificial intelligence system are implemented. Compared with the prior art, the beneficial effects of the electronic device provided by the present invention are the same as the beneficial effects of the soil deep loosening and tillage method based on an artificial intelligence system described in the above-mentioned technical solution, and will not be elaborated here.
[0091] The present invention also provides a computer-readable storage medium having a computer program stored thereon, characterized in that when the computer program is executed by a processor, the steps in the above-mentioned soil deep loosening and tillage method based on an artificial intelligence system are implemented. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided by the present invention are the same as the beneficial effects of the soil deep loosening and tillage method based on an artificial intelligence system described in the above technical solution, and will not be repeated here.
[0092] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any technical solution that can be easily conceived by a person skilled in the art within the technical scope disclosed in the present invention should be included in the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be based on the scope of protection of the claims.
Claims
1. A soil deep tillage method based on an artificial intelligence system, characterized in that: include: Step 1: Dig soil samples from different types of cultivated land and obtain visible light images of the soil samples; Step 2: Calculate the brightness component of the visible light image, and perform detail enhancement of the visible light image based on the brightness component to obtain a visible light image after detail enhancement; Step 3: Use convolutional layers and pooling layers to extract the overall and local features of the visible light image; Step 4: Combine the overall features and local features to obtain global features, and input the global features into the deep learning network for training to obtain a prediction model for cultivated land soil type; Step 5: using the cultivated land soil type prediction model to determine the soil type in the target cultivated land; Step 6: Determine the soil deep tillage method according to the soil type in the target farmland.
2. The soil deep tillage method based on an artificial intelligence system according to claim 1, characterized in that: The step 2: calculating the brightness component of the visible light image, and performing detail enhancement of the visible light image based on the brightness component to obtain a visible light image after detail enhancement, comprises: Step 2.1: Determine the brightness component of the visible light image based on the red, green and blue color channel values of the visible light image; the brightness component calculation formula is: Y=0.299R+0.587G+0.114B Among them, Y represents the brightness component, R represents the red color channel value, G represents the green color channel value, and B represents the blue color channel value; Step 2.2: performing logarithmic transformation on the brightness component to obtain a brightness logarithmic output; Step 2.3: Smoothing the visible light image based on the brightness logarithmic output to obtain a smoothed enhanced image; Step 2.4: Introduce contrast parameters to perform global detail enhancement on the smoothed enhanced image to obtain a detail-enhanced visible light image.
3. The soil deep tillage method based on an artificial intelligence system according to claim 2, characterized in that: The step 2.2: performing logarithmic transformation on the brightness component to obtain a brightness logarithmic output, comprises: Using the formula: The brightness component is logarithmically transformed to obtain a brightness logarithmic output; wherein L(x,y) represents the brightness component at the (x,y)th point, M is the maximum value of the pixel range of the image, and L g (x,y) represents the logarithmic output of the brightness at the (x,y)th point.
4. The soil deep tillage method based on an artificial intelligence system according to claim 3, characterized in that: The step 2.3: performing smoothing processing on the visible light image based on the brightness logarithmic output to obtain a smoothed enhanced image, comprises: An image smoothing enhancement formula is constructed based on a guided filter, and the image smoothing enhancement formula is used to smooth the visible light image to obtain a smoothed enhanced image; wherein the image smoothing enhancement formula is: Among them, L y (x, y) represents the image after smooth enhancement, H g (x,y) represents the guided filter, (ξ x ,ξ y ) represents the pixel coordinates of the domain, ω(x,y) represents the square area with a side length of r centered at the pixel point (x,y), |ω represents the number of pixels within ω(x,y), and a(ξ x ,ξ y ) represents the first adjustment coefficient b(ξ x ,x y )=μ(ξ x ,x y )-a(ξ x ,x y )μ(ξ x ,x y ),μ2(ξ x ,x y ) displayed The mean of the squared logarithmic outputs of brightness within ω(x,y), μ(ξ x ,ξ y ) represents the mean value of the logarithmic output of brightness within ω(x,y), σ 2 (ξ x ,ξ y ) represents the variance of the logarithmic output of brightness within ω(x,y), and ε represents the constant coefficient.
5. The soil deep tillage method based on an artificial intelligence system according to claim 4, characterized in that: In step 2.4, the formula is used: Perform global detail enhancement on the smoothed enhanced image to obtain a detail-enhanced visible light image; where α(x, y) represents the contrast parameter, η represents the parameter for image enhancement control, and L gmax represents the maximum value of the logarithmic output of brightness, λ represents the empirical parameter, represents the mean of the global logarithmic transformation, δ is a constant, and N represents the total number of pixels in the image.
6. The soil deep tillage method based on an artificial intelligence system according to claim 5, characterized in that: In step 3, the visible light image is sequentially input into the convolution layer and the pooling layer to obtain the overall feature; wherein the overall feature extraction formula is: c i =f(ω·H+b) In the formula, f is the activation function, ω is the convolution kernel, h is the convolution kernel size, b is the bias value, c is i is the feature vector extracted by the i-th convolutional layer, It represents the value of the features extracted by three different convolution kernels after the maximum pooling layer, and C represents the overall features extracted.
7. The soil deep tillage method based on an artificial intelligence system according to claim 6, characterized in that: In step 3, the local feature calculation formula is: Among them, F represents a visible light image, F1 represents the first feature map, F2 represents the second feature map, F3 represents the third feature map, F4 represents the fourth feature map, F″ represents a local feature, Conv1, Conv3, Conv4 and Conv6 represent 1×1×1 convolutions, Conv2 and Conv5 represent 3×3×3 convolutions, Cat represents a concatenation operation, and Maxpoo represents maximum pooling.
8. The soil deep tillage method based on artificial intelligence system according to claim 7, characterized in that: In step 4, the deep learning network is optimized and trained using the loss function to obtain a prediction model for cultivated land soil type; wherein the loss function is: in, represents the true category of soil type, m represents the total number of soil type categories, λ represents the regularization parameter, and y k represents the prediction result output by the deep learning network, and θ represents the network parameters.
9. An electronic device, comprising a bus, a transceiver, a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the transceiver, the memory, and the processor are connected via the bus, wherein: When the computer program is executed by the processor, the steps in the soil deep tillage method based on an artificial intelligence system as described in any one of claims 1 to 8 are implemented.
10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps in a soil deep tillage method based on an artificial intelligence system as described in any one of claims 1 to 8 are implemented.
Citation Information
Patent Citations
Image enhancing method based on self-adaptive block channel stretching
CN105809643A
Image enhancement method and system
CN107730475A
An image enhancement method in a dark light environment
CN109685742A
Low-illumination image enhancement method based on improved Retinex and logarithm image processing
CN110298796A
License plate image brightness processing method, device and equipment
CN110473158A
