Electric power image wireless transmission method and system based on semantic generality extraction, and storage medium

Abstract

The invention discloses an electric power image wireless transmission method based on semantic generality extraction. The method comprises the following steps: acquiring a to-be-transmitted target electric power image; carrying out blocking operation on the target power image; based on a pre-trained deep learning network, semantic feature information extraction is carried out on the segmented target power image; according to the semantic feature information and a clustering analysis scheme, performing common information extraction on the partitioned target power image to generate a codebook; the sending end sends data information and a codebook of a to-be-transmitted target power image to a receiving end; and the receiving end reconstructs the target power image according to the received data information so as to complete wireless transmission of the target power image. The invention also discloses a system for realizing the electric power image wireless transmission method based on semantic generality extraction, and a storage medium comprising the electric power image wireless transmission method based on semantic generality extraction. According to the method, the target power image is subjected to blocking, feature extraction and codebook generation, wireless transmission of the power image is achieved, and the method is high in reliability, good in accuracy and high in efficiency.

Description

Technical Field

[0001] The present invention belongs to the field of electrical automation, and particularly relates to a wireless transmission method, system and storage medium for power images based on semantic commonality extraction. Background Art

[0002] With the development of economic technology and the improvement of people's living standards, electric energy has become an essential secondary energy source in people's production and life, bringing endless convenience to people's production and life. Therefore, ensuring the stable and reliable supply of electric energy has become one of the most important tasks of the power system. At present, with the rapid development of Internet of Things technology and artificial intelligence technology, the wireless transmission of power system images is of great significance for the monitoring, identification, prediction, etc. of the power system.

[0003] In the process of wireless transmission of power images, it is generally necessary to compress the power images to improve the transmission efficiency. The early image compression technology was centered around the discrete cosine transform (DCT), and the typical representative was the JPEG standard: it achieved image compression through block transformation and quantization; however, this type of scheme was prone to block effects and detail loss at high compression ratios, and it was difficult to meet the requirements of high-precision image transmission. In recent years, the JPEG2000 technology based on wavelet transform has improved the compression efficiency through multi-resolution analysis and supports progressive transmission; however, this type of scheme has high computational complexity and poor real-time performance, and it is difficult to be applied to resource-constrained mobile terminals and low-bandwidth network environments. Summary of the Invention

[0004] One object of the present invention is to provide a wireless transmission method for power images based on semantic commonality extraction with high reliability, good accuracy and high efficiency.

[0005] Another object of the present invention is to provide a system for implementing the wireless transmission method for power images based on semantic commonality extraction.

[0006] A third object of the present invention is to provide a storage medium on which a computer program is stored; when the computer program is executed by a processor, the wireless transmission method for power images based on semantic commonality extraction is implemented.

[0007] The wireless transmission method for power images based on semantic commonality extraction provided by the present invention includes the following steps:

[0008] S1. Obtain the target power image to be transmitted;

[0009] S2. Perform a block operation on the target power image obtained in step S1;

[0010] S3. Extract semantic feature information from the block-divided target power image based on a pre-trained deep learning network;

[0011] S4. Extract common information from the segmented target power image according to the semantic feature information and clustering analysis scheme obtained in step S3 to generate a codebook;

[0012] S5. The sender sends the data information of the target power image to be transmitted and the codebook to the receiver;

[0013] S6. The receiver reconstructs the target power image according to the received data information to complete the wireless transmission of the target power image.

[0014] The block operation on the target power image obtained in step S1 described in step S2 specifically includes the following steps:

[0015] The target power image obtained in step S1 is denoted as I, I ∈ R H×W×3 ; H is the height of the target power image, and W is the width of the target power image;

[0016] Decompose the target power image I into sub-images of size B×B, where B is the side length of the sub-image: If H×W is divisible by B×B, directly decompose the target power image I; if H×W is not divisible by B×B, then pad the target power image I to I' ∈ R H'×W'×3 , H' is the height of the padded image, and W' is the width of the padded image, to ensure that H'×W' is divisible by B×B, and then decompose the padded target power image;

[0017] Finally, the decomposed sub-images are {B1, B2,..., B N}, and N is the total number of decomposed sub-images.

[0018] The extraction of semantic feature information from the segmented target power image based on the pre-trained deep learning network described in step S3 specifically includes the following steps:

[0019] Use the pre-trained residual neural network as the pre-trained deep learning network; among them, remove the last layer of the pre-trained residual neural network;

[0020] Input the decomposed sub-images {B1, B2,..., B N} into the pre-trained residual neural network in sequence for semantic feature extraction to obtain semantic feature vectors {f1, f2,..., f N}, where f i is the semantic feature vector extracted from the decomposed sub-image B i .

[0021] According to the semantic feature information and clustering analysis scheme obtained in step S3, extract common information from the segmented target power image to generate a codebook, which specifically includes the following steps:

[0022] Randomly select K features as the initial semantic feature clustering centers

[0023] For each feature vector f i Classification clustering label For where is the k-th semantic feature clustering center at the (t - 1)-th iteration, and || || is the L2 norm;

[0024] Update the clustering centers For where is the index set of all feature vectors assigned to the k-th cluster at the t-th iteration and represents the set the number of elements in, is the classification clustering label for the t-th iteration;

[0025] If the condition is satisfied, then stop the iterative update of the clustering centers; ε is the set threshold;

[0026] Finally, obtain K clustering centers T is the number of the final iteration;

[0027] For each feature clustering cluster k, find the corresponding segmented subgraph according to all the feature vectors within the feature clustering cluster, and calculate the pixel average value C k For where B ii is the pixel value of the segmented subgraph within the k-th feature clustering cluster, and |B k | is the number of segmented subgraphs within the k-th feature clustering cluster;

[0028] Form the K pixel average values into a codebook C as C = {C1, C2,..., C K}.

[0029] The sender in step S5 sends the data information and codebook of the target power image to be transmitted to the receiver, which specifically includes the following steps:

[0030] Perform channel coding on the position information vector of all segmented subgraphs in the original image, the clustering information vector to which the segmented subgraph belongs, and the codebook information, and send them to the receiver through wireless transmission.

[0031] According to the data information received, the receiving end reconstructs the target power image to complete the wireless transmission of the target power image, which specifically includes the following steps:

[0032] According to the received data information, the received codebook, the position information vector of all sub-block images in the original image, and the clustering information vector to which the sub-block image belongs, the receiving end replaces the position where the sub-block image is located with the codebook of the cluster to which it belongs, so as to complete the reconstruction of the target power image, and thus complete the wireless transmission of the target power image.

[0033] The present invention also provides a system for implementing the wireless transmission method of power images based on semantic commonality extraction, including an image acquisition module, an image segmentation module, a feature extraction module, a codebook generation module, an image transmission module, and an image reconstruction module; the image acquisition module, the image segmentation module, the feature extraction module, the codebook generation module, the image transmission module, and the image reconstruction module are connected in series in sequence; the image acquisition module is used to acquire the target power image to be transmitted and upload the data information to the image segmentation module; the image segmentation module is used to perform a segmentation operation on the acquired target power image according to the received data information and upload the data information to the feature extraction module; the feature extraction module is used to extract semantic feature information from the segmented target power image based on a pre-trained deep learning network according to the received data information and upload the data information to the codebook generation module; the codebook generation module is used to extract common information from the segmented target power image according to the received data information, the obtained semantic feature information, and the clustering analysis scheme to generate a codebook and upload the data information to the image transmission module; the image transmission module is used to send the data information and the codebook of the target power image to be transmitted by the sending end to the receiving end according to the received data information and upload the data information to the image reconstruction module; the image reconstruction module is used to reconstruct the target power image according to the received data information by the receiving end to complete the wireless transmission of the target power image.

[0034] The present invention also provides a storage medium, on which a computer program is stored; when the computer program is executed by a processor, the wireless transmission method of power images based on semantic commonality extraction is implemented.

[0035] The wireless transmission method, system, and storage medium of power images based on semantic commonality extraction provided by the present invention not only realize the wireless transmission of power images, but also have higher reliability, better accuracy, and higher efficiency by segmenting, feature extracting, and codebook generating the target power image. Description of the Drawings

[0036] Figure 1 It is a schematic flowchart of the method of the present invention.

[0037] Figure 2 It is a schematic diagram of the functional modules of the system of the present invention. Detailed implementation manners

[0038] As Figure 1 shown is a schematic diagram of the method flow of the method of the present invention: The method for wireless transmission of power images based on semantic commonality extraction disclosed in the present invention includes the following steps:

[0039] S1. Obtain the target power image to be transmitted;

[0040] S2. Perform a blocking operation on the target power image obtained in step S1; specifically, it includes the following steps:

[0041] The target power image obtained in step S1 is represented as I, I ∈ R H×W×3 ; H is the height of the target power image, and W is the width of the target power image;

[0042] Decompose the target power image I into sub-images of size B×B, where B is the side length of the sub-image: If H×W can be divided evenly by B×B, there is no need for padding, and directly decompose the target power image I; if H×W cannot be divided evenly by B×B, then pad the target power image I to I' ∈ R H'×W'×3 , H' is the height of the padded image, and W' is the width of the padded image, to ensure that H'×W' can be divided evenly by B×B, and then decompose the padded target power image;

[0043] Finally, the decomposed sub-images are {B1, B2,..., B N}, and N is the total number of the decomposed sub-images;

[0044] S3. Based on the pre-trained deep learning network, extract semantic feature information from the blocked target power image; specifically, it includes the following steps:

[0045] Adopt the pre-trained residual neural network as the pre-trained deep learning network; among them, remove the last layer of the pre-trained residual neural network;

[0046] Input the decomposed sub-images {B1, B2,..., B N} into the pre-trained residual neural network in sequence for semantic feature extraction, and obtain semantic feature vectors {f1, f2,..., f N}, where f i is the semantic feature vector extracted from the decomposed sub-image B i ;

[0047] In specific implementation, the sub-images input into the pre-trained residual neural network need to be normalized in advance;

[0048] S4. Extract common information from the segmented target power image according to the semantic feature information and clustering analysis scheme obtained in step S3 to generate a codebook. The specific steps are as follows:

[0049] Randomly select K features as the initial semantic feature clustering centers

[0050] For each feature vector f i Classification clustering label For Where Is the k-th semantic feature clustering center at the (t - 1)-th iteration, and || || is the L2 norm;

[0051] Update the clustering center For Where Is the index set of all feature vectors assigned to the k-th cluster at the t-th iteration and Represents the set The number of elements in, Is the classification clustering label for the t-th iteration;

[0052] If the condition Is satisfied, stop the iterative update of the clustering center; ε is the set threshold;

[0053] Finally, obtain K clustering centers T is the number of times of the final iteration;

[0054] For each feature clustering cluster k, find the corresponding segmented subgraph according to all feature vectors within the feature clustering cluster, and calculate the pixel average value C k For Where B ii Is the pixel value of the segmented subgraph within the k-th feature clustering cluster, |B k | Is the number of segmented subgraphs within the k-th feature clustering cluster;

[0055] Form the K pixel average values into a codebook C as C = {C1, C2,..., C K};

[0056] S5. The sending end sends the data information and codebook of the target power image to be transmitted to the receiving end. The specific steps are as follows:

[0057] Perform channel coding (preferably using LDPC channel coding) on the position information vector of all segmented subgraphs in the original image, the clustering information vector to which the segmented subgraph belongs, and the codebook information, and send them to the receiving end through wireless transmission;

[0058] Among them, the position information vector of all sub-block images in the original image and the clustering information vector to which the sub-block images belong are preferably transmitted with high reliability to ensure error-free reception at the receiving end;

[0059] S6. The receiving end reconstructs the target power image based on the received data information to complete the wireless transmission of the target power image; specifically, it includes the following steps:

[0060] The receiving end replaces the positions where the sub-block images are located with the codebooks of the clusters to which they belong according to the received data information, the received codebook, the position information vector of all sub-block images in the original image, and the clustering information vector of the sub-block images, so as to complete the reconstruction of the target power image, thereby completing the wireless transmission of the target power image.

[0061] The method of the present invention extracts semantic information based on deep learning and extracts its common features through clustering, reducing the frequency band resources consumed during transmission; at the same time, it solves the redundant transmission of similar content, reduces the data transmission volume, and ensures the safe and stable operation of power grid service data.

[0062] As Figure 2 Shown is a schematic diagram of the functional modules of the system of the present invention: The system for implementing the wireless transmission method of power images based on semantic common feature extraction disclosed in the present invention includes an image acquisition module, an image segmentation module, a feature extraction module, a codebook generation module, an image transmission module, and an image reconstruction module; the image acquisition module, the image segmentation module, the feature extraction module, the codebook generation module, the image transmission module, and the image reconstruction module are connected in series in sequence; the image acquisition module is used to acquire the target power image to be transmitted and upload the data information to the image segmentation module; the image segmentation module is used to perform a segmentation operation on the acquired target power image according to the received data information and upload the data information to the feature extraction module; the feature extraction module is used to extract semantic feature information from the segmented target power image based on a pre-trained deep learning network according to the received data information and upload the data information to the codebook generation module; the codebook generation module is used to extract common information from the segmented target power image according to the received data information, the obtained semantic feature information, and the clustering analysis scheme to generate a codebook and upload the data information to the image transmission module; the image transmission module is used to send the data information and the codebook of the target power image to be transmitted by the sending end to the receiving end according to the received data information and upload the data information to the image reconstruction module; the image reconstruction module is used to reconstruct the target power image by the receiving end according to the received data information to complete the wireless transmission of the target power image.

[0063] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, a system, or a computer program product. Therefore, the present application can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code. The solutions in the embodiments of the present application can be implemented in various computer languages. For example, object-oriented programming languages such as Java and interpreted scripting languages such as JavaScript, etc.

[0064] The present application is described with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to the embodiments of the present application. It should be understood that each flow and / or block in the flowchart and / or block diagram, as well as the combination of flows and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate a device for implementing the functions specified in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0065] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory generate a manufactured article including an instruction device, and the instruction device implements the functions specified in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0066] These computer program instructions can also be loaded onto a computer or other programmable data processing device, so that a series of operation steps are executed on the computer or other programmable device to generate a computer-implemented process. Thus, the instructions executed on the computer or other programmable device provide steps for implementing the functions specified in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0067] Although the preferred embodiments of the present application have been described, those skilled in the art can make additional changes and modifications once they learn the basic creative concepts. Therefore, the appended claims are intended to be construed to include the preferred embodiments as well as all changes and modifications falling within the scope of the present application.

[0068] Obviously, those skilled in the art can make various changes and modifications to this application without departing from the spirit and scope of this application. Thus, if these modifications and variations of this application fall within the scope of the claims of this application and their equivalent technologies, this application is also intended to cover these modifications and variations.

Claims

1. A wireless transmission method for power images based on semantic commonality extraction, comprising the following steps: S1. Obtain the target power image to be transmitted; S2. Perform a blocking operation on the target power image obtained in step S1; S3. Based on a pre-trained deep learning network, extract semantic feature information from the blocked target power image; S4. According to the semantic feature information obtained in step S3 and the clustering analysis scheme, extract common information from the blocked target power image to generate a codebook; S5. The sender sends the data information of the target power image to be transmitted and the codebook to the receiver; S6. The receiver reconstructs the target power image according to the received data information to complete the wireless transmission of the target power image.

2. The method for wireless transmission of power images based on semantic commonality extraction according to claim 1, wherein The blocking operation on the target power image obtained in step S1 described in step S2 specifically includes the following steps: The target power image obtained in step S1 is denoted as I, I ∈ R H×W×3 ; H is the height of the target power image, and W is the width of the target power image; Decompose the target power image I into sub-images of size B×B, where B is the side length of the sub-image: if H×W is divisible by B×B, directly decompose the target power image I; If \(H\times W\) cannot be divided evenly by \(B\times B\), then the target power image \(I\) is padded to \(I'\in\mathbb{R}\) H'×W'×3 , where \(H'\) is the height of the padded image and \(W'\) is the width of the padded image, to ensure that \(H'\times W'\) can be divided evenly by \(B\times B\), and then the padded target power image is decomposed; Finally , the decomposed subgraphs obtained are {B1, B2,..., B N}, where N is the total number of the decomposed subgraphs.

3. The method for wireless transmission of power images based on semantic commonality extraction according to claim 2, characterized in that The extraction of semantic feature information from the blocked target power image based on the pre-trained deep learning network described in step S3 specifically includes the following steps: Use a pre-trained residual neural network as the pre-trained deep learning network; among them, remove the last layer of the pre-trained residual neural network; The decomposed subgraphs {B1, B2,..., B N} are sequentially input into a pre-trained residual neural network for semantic feature extraction to obtain semantic feature vectors {f1, f2,..., f N}, where f i is the semantic feature vector extracted from the decomposed subgraph B i .

4. The method for wireless transmission of power images based on semantic commonality extraction according to claim 3, characterized in that The extraction of common information from the blocked target power image according to the semantic feature information obtained in step S3 and the clustering analysis scheme to generate a codebook described in step S4 specifically includes the following steps: Randomly select K features as the initial semantic feature clustering centers For each feature vector f i Classification clustering label is where is the k-th semantic feature clustering center at the (t - 1)-th iteration, and || || is the L2 norm; Update the cluster center For Where is the index set of all feature vectors assigned to the k-th cluster at the t-th iteration and denotes the set the number of elements in, is the classification cluster label after t iterations; If the condition is satisfied then stop the iterative update of the cluster centers; ε is the set threshold value; Finally, K clustering centers are obtained. T is the number of times of the final iteration; For each feature clustering cluster k, find the corresponding sub-block graph according to all the feature vectors within the feature clustering cluster, and calculate the pixel average value C k is where B ii is the pixel value of the sub-block graph within the k-th feature clustering cluster, |B k | is the number of sub-block graphs within the k-th feature clustering cluster; The average values of K pixels are used to form a codebook C, where C = {C1, C2,..., C K}.

5. The wireless power image transmission method based on semantic commonality extraction according to claim 4, characterized in that The sender described in step S5 sends the data information of the target power image to be transmitted and the codebook to the receiver, specifically including the following steps: Channel-encode the position information vector of all blocked sub-images in the original image, the clustering information vector to which the blocked sub-images belong, and the codebook information, and send them to the receiver by wireless transmission.

6. The method for wireless transmission of power images based on semantic commonality extraction according to claim 5, wherein The receiver described in step S6 reconstructs the target power image according to the received data information to complete the wireless transmission of the target power image, specifically including the following steps: The receiver replaces the position where the blocked sub-image is located with the codebook of the cluster to which it belongs according to the received data information, the received codebook, the position information vector of all blocked sub-images in the original image, and the clustering information vector to which the blocked sub-images belong, so as to complete the reconstruction of the target power image, thereby completing the wireless transmission of the target power image.

7. A system for implementing the power image wireless transmission method based on semantic commonality extraction according to any one of claims 1 to 6, characterized in that It includes an image acquisition module, an image blocking module, a feature extraction module, a codebook generation module, an image sending module, and an image reconstruction module; the image acquisition module, the image blocking module, the feature extraction module, the codebook generation module, the image sending module, and the image reconstruction module are connected in series in sequence; the image acquisition module is used to obtain the target power image to be transmitted and upload the data information to the image blocking module; The image blocking module is used to perform a blocking operation on the obtained target power image according to the received data information and upload the data information to the feature extraction module; The feature extraction module is used to extract semantic feature information from the segmented target power image based on the received data information and a pre-trained deep learning network, and upload the data information to the codebook generation module; The codebook generation module is used to extract common information from the segmented target power image according to the received data information, the obtained semantic feature information and the clustering analysis scheme to generate a codebook, and upload the data information to the image sending module; The image sending module is used to send the data information and the codebook of the target power image to be transmitted to the receiving end according to the received data information, and upload the data information to the image reconstruction module; The image reconstruction module is used to reconstruct the target power image at the receiving end according to the received data information to complete the wireless transmission of the target power image.

8. A storage medium, on which a computer program is stored; when the computer program is executed by a processor, the wireless power image transmission method based on semantic commonality extraction according to any one of claims 1 to 6 is implemented.

Images