Geological surveying and mapping image enhancement processing method

Abstract

The invention provides a geological surveying and mapping image enhancement processing method, which belongs to the field of image enhancement, and comprises a plurality of key modules: an image acquisition module acquires image data of a geological area through high-precision equipment; the primary image enhancement module extracts geological features of different scales by using a multi-scale convolution kernel, and performs composite normalization to enhance image details; the secondary image enhancement module adopts a variational auto-encoder to optimize image details, improve image definition and contrast and enhance details of a low-contrast area; the tertiary image enhancement module enhances the edge features of the image through adaptive filtering and improves the visibility of important geological features; the quartic image enhancement module uses a Gabor filter to extract image texture features, and the recognizability of geologic features is enhanced. All the modules work cooperatively, the quality of geological images is improved, geological features are clearer and easier to recognize, the system is widely applied to the fields of geological exploration and mineral resource investigation, and efficient technical support is provided for geological analysis.

Description

Technical Field

[0001] The invention belongs to the field of image enhancement, and in particular relates to a geological mapping image enhancement processing method. Background Art

[0002] Geological mapping images play a vital role in geological exploration, mineral resource development and environmental monitoring. Through image acquisition, geological experts can obtain information about geological structure, rock formations, mineral distribution, etc. These images provide an important basis for geological analysis; with the advancement of image acquisition technology, the collected geological images usually contain more complex textures, edges and details. How to extract and enhance this information has become the key to improving the accuracy and quality of image analysis; therefore, the development of new image enhancement technology, especially processing methods that can optimize image details, improve image clarity, and enhance the recognizability of geological features, has become the key to improving the quality of geological mapping images and improving exploration efficiency.

[0003] By combining modern image enhancement techniques based on convolutional neural networks, variational autoencoders, Gabor filters, etc., the details and clarity of geological images can be enhanced, and the ability to recognize geological features can be improved. Summary of the invention

[0004] The present invention provides a geological mapping image enhancement processing method, which aims to improve the quality of geological mapping images and the recognition accuracy of geological features through image processing and deep learning technology. The method uses multi-module image enhancement processing, including multi-scale image enhancement, image detail optimization, edge enhancement, texture feature extraction and other technologies to improve the clarity and recognizability of geological mapping images.

[0005] The present invention aims to propose a geological mapping image enhancement processing model and provide a geological mapping image enhancement processing method, which includes the following steps.

[0006] S1. Geological surveying and mapping image acquisition: Geological surveying and mapping images are collected through data acquisition equipment.

[0007] S2. Construct an image enhancement module, use two convolution kernels of different sizes, process geological features of multiple sizes in the image, and composite and normalize the geological features of different sizes.

[0008] S3. Construct a secondary image enhancement module and use variational autoencoder to optimize image details and enhance the clarity of feature details of geological mapping images.

[0009] S4. Construct a three-dimensional image enhancement module to perform adaptive filtering enhancement on the edge area of ​​the geological mapping image, and dynamically adjust the filtering parameters according to the local characteristics of the edge area to enhance the edge details.

[0010] S5. Construct a quartic image enhancement module, use Gabor spatial frequency filter to enhance the geological mapping image, extract the texture features in the enhanced geological mapping image, and improve the recognizability of geological features.

[0011] Preferably, in step S2, for the primary image enhancement module, the input geological mapping image is I(x, y), where (x, y) is the coordinate of the image, and I(x, y) is the pixel value of the image at that point. Two convolution kernels of different sizes are constructed to perform convolution operations on the images respectively. The formula of the convolution operation is: Where K(i, j) is the value of the convolution kernel, (2m+1)×(2n+1) is the size of the convolution kernel, (I*K)(x, y) is the output geological mapping image after the convolution operation, and the geological mapping images processed by two convolution kernels of different sizes are fused by weighted averaging. fused (x,y))=α·(I1*K1)(x,y)+(1-α)·(I2*K2)(x,y) where I1 and I2 are geological mapping images processed using two different convolution kernels, and α is the weighting coefficient.

[0012] Preferably, in step S2, for a primary image enhancement module, multi-scale convolution processing is performed to extract details of geological features of different sizes in the image, ensuring that geological information of different scales is accurately retained, and the contrast of features is enhanced by composite normalization, so that both subtle small features and large geological structures are effectively improved and optimized.

[0013] Preferably, in step S3, for the secondary image enhancement module, the encoder q(z|I input ) Input geological mapping image I input Mapped to the latent space, the generated latent variable z distribution is a Gaussian distribution Where μ(I input ) is the mean of the latent variables output by the encoder, σ(I input ) is the standard deviation of the latent variable output by the encoder, and then passed through the decoder p(I output |z) encodes the input geological mapping image to obtain the mean μ and standard deviation σ in the latent space, z = μ(I input )+σ(I input )·∈where μ(I input ) is the mean of the latent variables output by the encoder, σ(I input ) is the standard deviation of the latent variable output by the encoder, ∈ is the noise sampled from the standard normal distribution, and the reconstructed geological mapping image I is generated by inputting the latent variable z output =Decoder(z), where I output The geological mapping image is generated by the decoder, and Decoder is the output layer of the neural network.

[0014] Preferably, in step S3, for the secondary image enhancement module, after mapping the image to the latent space using the variational autoencoder, the image is reconstructed through the decoder to optimize the details in the image, further enhance the image details, especially the clarity of the geological features, so that the details are more prominent, thereby improving the image quality and recognizability.

[0015] Preferably, in step S4, for the three-time image enhancement module, as Figure 2 As shown in the figure, the Sobel edge detection algorithm is used to detect the edge area of ​​the geological mapping image, and the Sobel operator is used to calculate the gradient of the geological mapping image. Among them, G x For horizontal gradients, an adaptive filter is applied to the edge area, changing its filtering strength according to the local characteristics of the signal in is the gradient of a single pixel of a geological mapping image, β is the adjustment coefficient of the adaptive filter, I enhanced (x, y) is the details of the processed geological mapping image.

[0016] Preferably, in step S4, for the tertiary image enhancement module, the edge portion of the image is enhanced through adaptive filtering technology, so that the edge portion of the image becomes clearer and the boundary of the geological feature becomes more distinct, which is helpful for subsequent analysis and judgment of the geological structure and improves the analysis value of the image.

[0017] Preferably, in step S5, for the quartic image enhancement module, a Gabor spatial frequency filter is used to extract the texture features of the geological mapping image. Where G(x, y, f0, θ, γ) is the output of the Gabor spatial frequency filter, f0 is the frequency parameter of the filter, θ is the direction parameter of the filter, and γ is the standard deviation of the Gaussian part.

[0018] Preferably, in step S5, for the quartic image enhancement module, a Gabor spatial frequency filter is used to extract texture features of the image, and the subtle texture in the image is captured by filtering at different frequencies and directions, so that the geological structure and mineral distribution in the image are more obvious, which is convenient for subsequent feature recognition and analysis.

[0019] Compared with the prior art, the present invention has the following technical effects: The present invention combines the advantages of color image processing and geological feature extraction, effectively retains the geological feature information in the image through multiple image enhancement modules, and captures the key geological features in the image through detail enhancement, edge optimization and texture extraction, etc., providing high-quality input data for subsequent analysis and feature extraction, significantly improving the recognition accuracy of geological features, and improving the efficiency and accuracy of geological mapping image processing through the collaborative work of image enhancement modules. The ultimately generated geological image has the characteristics of high precision, rich details and real-time processing capabilities, and can be widely used in geological exploration, mineral resource investigation, environmental monitoring and other fields, improving the accuracy and efficiency of geological image analysis. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 It is a flow chart of a geological surveying and mapping image enhancement processing method provided by the present invention.

[0021] Figure 2 It is a structural diagram of the secondary image enhancement module provided by the present invention.

[0022] Figure 3 The left half is a low-resolution geological mapping image provided by the present invention. Figure 3 The right half is the geological mapping image effect after image enhancement processing. DETAILED DESCRIPTION

[0023] The invention proposes a geological mapping image enhancement processing method, which realizes efficient image enhancement and geological feature optimization through the collaborative work of multiple key modules; the method comprises: a geological mapping image acquisition module, which acquires multi-band images through high-precision equipment to ensure data quality; a primary image enhancement module, which extracts multi-scale geological features by using convolution kernels of different sizes, and performs composite normalization to enhance the image detail level; a secondary image enhancement module, which optimizes low-contrast areas and noise areas by using variational autoencoders to improve image clarity; a tertiary image enhancement module, which enhances edge features by using adaptive filtering to make rock strata and crack boundaries clearer; a quaternary image enhancement module, which extracts texture features by using Gabor filtering to optimize geological structure information; the images generated by the invention are rich in details and clear in features, suitable for geological exploration and resource analysis, and improve the efficiency and accuracy of geological image processing.

[0024] See also Figure 1 As shown, a geological mapping image enhancement processing method in an embodiment of the present application has the following process.

[0025] S1. Geological mapping image acquisition: Geological mapping images are collected through geological mapping image data acquisition equipment, and measurement unit data is obtained at the same time. 4,000 images are obtained, and the collected geological mapping images are manually annotated. The obtained images constitute a geological mapping image dataset. The dataset is divided into a training set and a validation set in a ratio of 7:3.

[0026] S2. Construct an image enhancement module, use two convolution kernels of different sizes, process geological features of multiple sizes in the image, and composite and normalize the geological features of different sizes.

[0027] Further, in step S2, for the primary image enhancement module, the input geological mapping image is I(x, y), where (x, y) is the coordinate of the image, and I(x, y) is the pixel value of the image at that point. Two convolution kernels of different sizes are constructed to perform convolution operations on the images respectively. The formula of the convolution operation is: Where K(i, j) is the value of the convolution kernel, (2m+1)×(2n+1) is the size of the convolution kernel, (I*K)(x, y) is the output geological mapping image after the convolution operation, and the geological mapping images processed by two convolution kernels of different sizes are fused by weighted averaging. fused (x,y)=α·(I1*K1)(x,y)+(1-α)·(I2*K2)(x,y) where I1 and I2 are geological mapping images processed using two different convolution kernels, and α is the weighting coefficient.

[0028] S3. Construct a secondary image enhancement module and use variational autoencoder to optimize image details and enhance the clarity of feature details of geological mapping images.

[0029] Further, in step S3, for the secondary image enhancement module, such as Figure 2 As shown, through the encoder Q(z|I input ) Input geological mapping image I input Mapped to the latent space, the generated latent variable z distribution is a Gaussian distribution Where μ(I input ) is the mean of the latent variables output by the encoder, σ(I input ) is the standard deviation of the latent variable output by the encoder, and then passed through the decoder p(I output |z) encodes the input geological mapping image to obtain the mean μ and standard deviation σ in the latent space, z = μ(I input )+σ(I input )·∈, where μ(I input ) is the mean of the latent variables output by the encoder, σ(I input) is the standard deviation of the latent variable output by the encoder, ∈ is the noise sampled from the standard normal distribution, and the reconstructed geological mapping image I is generated by inputting the latent variable z output =Decoder(z), where I output The geological mapping image is generated by the decoder, and Decoder is the output layer of the neural network.

[0030] Furthermore, in step S3, the code for the secondary image enhancement module is as follows: import tensorflow as tf from tensorflow.keras.layers import Input,Dense,Flatten,Reshape,Lambda from tensorflow.keras.models import Model from tensorflow.keras import backend as K def sampling(args): z_mean,z_log_var=args epsilon=K.random_normal(shape=K.shape(z_mean)) return z_mean+K.exp(0.5*z_log_var)*epsilon input_shape = (128,128,1) latent_dim=16 inputs=Input(shape=input_shape) x = Flatten()(inputs) x=Dense(256,activation='relu')(x) x=Dense(128,activation='relu')(x) z_mean=Dense(latent_dim)(x) z_log_var=Dense(latent_dim)(x) z=Lambda(sampling,output_shape=(latent_dim,))([z_mean,z_log_var]) encoder=Model(inputs,[z_mean,z_log_var,z]) latent_inputs=Input(shape=(latent_dim,)) x=Dense(128,activation='relu')(latent_inputs) x=Dense(256,activation='relu')(x) x=Dense(128*128,activation='sigmoid')(x) outputs=Reshape((128,128,1))(x) decoder=Model(latent_inputs,outputs) vae_outputs=decoder(encoder(inputs)[2]) vae=Model(inputs,vae_outputs) reconstruction_loss=tf.reduce_mean(tf.reduce_sum(tf.keras.losses.binary_crossentropy(inputs,vae_outputs),axis=(1,2))) kl_loss=-0.5*tf.reduce_sum(1+z_log_var-tf.square(z_mean)-tf.exp(z_log_var),axis=-1) vae_loss=tf.reduce_mean(reconstruction_loss+kl_loss) vae.add_loss(vae_loss) vae.compile(optimizer='adam').

[0031] S4. Construct a three-dimensional image enhancement module to perform adaptive filtering enhancement on the edge area of ​​the geological mapping image, and dynamically adjust the filtering parameters according to the local characteristics of the edge area to enhance the edge details.

[0032] Further, in step S4, for the third image enhancement module, the edge area of ​​the geological mapping image is detected using the Sobel edge detection algorithm, and the Sobel operator is used to calculate the gradient of the geological mapping image. Among them, G x For horizontal gradients, an adaptive filter is applied to the edge area, changing its filtering strength according to the local characteristics of the signal in is the gradient of a single pixel of a geological mapping image, β is the adjustment coefficient of the adaptive filter, I enhanced (x, y) is the details of the processed geological mapping image.

[0033] Furthermore, in step S4, the code for the three-time image enhancement module is as follows: def adaptive_edge_enhancement(image): gray=cv2.cvtColor(image,cv2.COLOR_BGR2GRAY) edges=cv2.Sobel(gray,cv2.CV_64F,1,1,ksize=3) edges_normalized=cv2.normalize(edges,None,0,255,cv2.NORM_MINMAX) enhanced_image=cv2.addWeighted(gray,1,edges_normalized.astype(np.uint8),0.5,0) return enhanced_image input_image=cv2.imread('input_image.jpg') output_image=adaptive_edge_enhancement(input_image) cv2.imwrite('output_image.jpg',output_image).

[0034] S5. Construct a quartic image enhancement module, use Gabor spatial frequency filter to enhance the geological mapping image, extract the texture features in the enhanced geological mapping image, and improve the recognizability of geological features.

[0035] Further, in step S5, for the quartic image enhancement module, a Gabor spatial frequency filter is used to extract the texture features of the geological mapping image. Where G(x, y, f0, θ, γ) is the output of the Gabor spatial frequency filter, f0 is the frequency parameter of the filter, θ is the direction parameter of the filter, and γ is the standard deviation of the Gaussian part.

[0036] Furthermore, in step S5, the code for the fourth image enhancement module is as follows: def gabor_filter_enhancement(image,kernel_size=31,sigma=4.0,theta=0,lambd=10.0,gamma=0.5,psi=0): gabor_kernel=cv2.getGaborKernel((kernel_size,kernel_size),sigma,theta,lambd,gamma,psi,ktype=cv2.CV_32F) filtered_image=cv2.filter2D(image,cv2.CV_8UC3,gabor_kernel) return filtered_image input_image=cv2.imread('input_image.jpg',cv2.IMREAD_GRAYSCALE) enhanced_image=gabor_filter_enhancement(input_image) cv2.imwrite('enhanced_image.jpg', enhanced_image).

[0037] Further, in step S5, if Figure 3 As shown, Figure 3 The left half shows a low-resolution geological mapping image. Figure 3 The right half shows the geological mapping image after image enhancement processing. It can be seen that the image clarity and detail performance have been improved, the details are presented more clearly, and subsequent identification is more accurate.

[0038] The above are only preferred embodiments of the present invention. It should be pointed out that a person skilled in the art can make several modifications and improvements without departing from the inventive concept of the present invention, and these all fall within the protection scope of the present invention.

Claims

1. A geological mapping image enhancement processing method, characterized in that: The following steps are involved: S1. Geological surveying and mapping image acquisition: collecting geological surveying and mapping images through data acquisition equipment; S2, construct an image enhancement module, use two convolution kernels of different sizes to process geological features of multiple sizes in the image, and normalize the geological features of different sizes; S3, construct a secondary image enhancement module, use variational autoencoder to optimize image details, and enhance the clarity of geological mapping image feature details; S4, constructing a three-dimensional image enhancement module to perform adaptive filtering enhancement on the edge area of ​​the geological mapping image, and dynamically adjusting the filtering parameters according to the local characteristics of the edge area to enhance the edge details; S5. Construct a quartic image enhancement module, use Gabor spatial frequency filter to enhance the geological mapping image, extract the texture features in the enhanced geological mapping image, and improve the recognizability of geological features.

2. A geological mapping image enhancement processing method according to claim 1, characterized in that: The specific method for constructing the primary image enhancement module in S2 is as follows: the input geological mapping image is I(x, y), where (x, y) is the coordinate of the image, and I(x, y) is the pixel value of the image at that point. Two convolution kernels of different sizes are constructed to perform convolution operations on the images respectively. The formula for the convolution operation is: Where K(i, j) is the value of the convolution kernel, (2m+1)×(2n+1) is the size of the convolution kernel, (I*K)(x, y) is the output geological mapping image after the convolution operation, and the geological mapping images processed by two convolution kernels of different sizes are fused by weighted averaging. fused (x, y) = α·(I1*K1)(x, y)+(1-α)·(I2*K2)(x, y), where I1 and I2 are geological mapping images processed using two different convolution kernels, and α is the weighting coefficient.

3. A geological mapping image enhancement processing method according to claim 1, characterized in that: The specific method of constructing the secondary image enhancement module in S3 is: through the encoder q(z|I input ) Input geological mapping image I input Mapped to the latent space, the generated latent variable z distribution is a Gaussian distribution Where μ(I input ) is the mean of the latent variables output by the encoder, μ(I input ) is the standard deviation of the latent variable output by the encoder, and then, through the decoder p(I output |z) encodes the input geological mapping image to obtain the mean μ and standard deviation σ in the latent space, z = μ(I input )+σ(I input )·∈where μ(I input ) is the mean of the latent variables output by the encoder, σ(I input ) is the standard deviation of the latent variable output by the encoder, ∈ is the noise sampled from the standard normal distribution, and the reconstructed geological mapping image I is generated by inputting the latent variable z output =Decoder(z), where I output It is the geological mapping image generated by the decoder, and Decoder is the output layer of the neural network.

4. A geological mapping image enhancement processing method according to claim 1, characterized in that: The specific method for constructing the three-dimensional image enhancement module in S4 is: using the Sobel edge detection algorithm to detect the edge area of ​​the geological mapping image, and the Sobel operator is used to calculate the gradient of the geological mapping image. Among them, G x For horizontal gradients, an adaptive filter is applied to the edge area, changing its filtering strength according to the local characteristics of the signal in is the gradient of a single pixel of a geological mapping image, β is the adjustment coefficient of the adaptive filter, I enhanced (x, y) is the details of the processed geological mapping image.

5. The method for enhancing geological mapping images according to claim 1, characterized in that: The specific method for constructing the quartic image enhancement module in S5 is: using Gabor spatial frequency filter to extract the texture features of geological mapping images Where G(x, y, f0, θ, γ) is the output of the Gabor spatial frequency filter, f0 is the frequency parameter of the filter, θ is the direction parameter of the filter, and γ is the standard deviation of the Gaussian part.

Images