Image transmission method and device based on artificial intelligence algorithm

Abstract

The invention discloses an image transmission method and device based on an artificial intelligence algorithm, and relates to the technical field of image transmission, and the device comprises an image collection module, an image decomposition module, an image transmission module, an intelligent optimization module and an interaction control module. By means of AI technologies such as deep learning and computer vision, the image content is automatically analyzed, the network condition is monitored, and the transmission strategy is dynamically optimized, so that the image transmission efficiency is improved, the data transmission loss is reduced, the adaptability to the network environment is enhanced, and the challenges of a traditional remote sensing image transmission method in the aspects of bandwidth, delay, packet loss rate and the like are solved. Through introduction of artificial intelligence, automatic management and intelligent optimization of image transmission are realized, manual intervention is reduced, and task execution efficiency and controllability of a transmission process are improved.

Description

Technical Field

[0001] This invention relates to the field of image transmission technology, and more specifically, to an image transmission method and apparatus based on artificial intelligence algorithms. Background Technology

[0002] Remote sensing images are a very powerful data source, widely used in environmental monitoring, agriculture, urban management, disaster monitoring and many other fields. Remote sensing images have the characteristics of wide coverage, multi-dimensional information, accurate monitoring and efficient acquisition, which greatly promotes the development of various fields. Whether in environmental protection, agricultural development, disaster response, or national defense and scientific research, remote sensing images play an irreplaceable role. Therefore, they have an undeniable importance in all aspects of modern society.

[0003] With the rapid development of remote sensing and drone technologies, the acquisition of remote sensing image data has become a key tool in modern scientific research, environmental monitoring, agricultural monitoring, and post-disaster assessment. These remote sensing images typically have high resolution, high data volume, and high complexity. However, during transmission and processing, these images often face technical challenges such as bandwidth limitations, network latency, and image distortion. In particular, when transmitting remote sensing images in real time, how to make full use of limited bandwidth and network resources while ensuring image quality and avoiding problems such as image loss and excessive latency is a core issue that remote sensing image transmission technology urgently needs to address. Summary of the Invention

[0004] The main objective of this invention is to provide an image transmission method and apparatus based on artificial intelligence algorithms to overcome the problems mentioned in the background art.

[0005] To achieve the above objectives, according to one aspect of the present invention, an image transmission method based on an artificial intelligence algorithm is provided, the method comprising the following steps:

[0006] Raw remote sensing image data is collected using remote sensing equipment, and preliminary data cleaning is performed on the remote sensing images.

[0007] The remote sensing image is then segmented, and the regional transmission value and image complexity of the remote sensing image are analyzed to obtain the overall image transmission value, which is then used to label the remote sensing image.

[0008] The overall performance of the equipment is comprehensively evaluated, and transmission tasks are matched accordingly. The specific process is as follows: All UAV parameters are acquired and normalized to the [0, 1] interval. The normalized UAV parameters are then processed to obtain the UAV's score. Several task types are set. When executing a transmission task, the parameter with the highest transmission occupancy rate is detected, and it is determined as a real-time transmission task, with its weight increased. The UAV parameters are compared and analyzed one by one with the standard threshold interval to determine the priority transmission queue for the UAVs. The overall image transmission value and the priority transmission queue are matched, and then the corresponding remote sensing image is transmitted.

[0009] In a preferred embodiment of the present invention, remote sensing images are segmented, and the region optimization values ​​and image complexity of the remote sensing images are analyzed, including:

[0010] Remote sensing images are segmented in real time using a deep learning model to obtain several sub-images;

[0011] Identify and label the corresponding type regions for several sub-images;

[0012] Determine the corresponding weight based on the selected task mode;

[0013] Based on the corresponding weights, the area superiority index of each sub-image is calculated. The area superiority indices of each sub-image in the remote sensing image are summed to obtain the area superiority transmission value of the remote sensing image. If the area superiority transmission value is less than the preset first threshold, the comprehensive transmission value is directly taken as the preset value.

[0014] If the area-optimized transmission value is greater than or equal to the preset first threshold, the remote sensing image will be converted into an 8-bit grayscale image.

[0015] Identify and count the number of times each pixel value appears in a grayscale image, calculate the probability corresponding to each pixel value, and calculate the image complexity using the information entropy formula.

[0016] The overall image transmission value of remote sensing images is obtained by analyzing the regional transmission value and image complexity of remote sensing images.

[0017] In a preferred embodiment of the present invention, the remote sensing image is labeled, including:

[0018] A preset image overall transmission threshold range is established, and the image overall transmission value is compared with the set image overall transmission threshold range.

[0019] When the overall image transmission value is greater than the maximum value of the set overall image transmission threshold range, the image is marked as the first priority transmission image;

[0020] When the overall image transmission value is within the set overall image transmission threshold range, the image is marked as the second priority transmission image;

[0021] When the overall image transmission value is less than the minimum value of the set overall image transmission threshold range, the image is marked as the third priority transmission image.

[0022] In a preferred embodiment of the present invention, the normalized UAV parameters are processed to obtain the UAV's score value, including:

[0023] The drone's endurance, bandwidth, computing power, latency, packet loss, and jitter values ​​are input into a linear model to obtain the drone's score.

[0024] In a preferred embodiment of the present invention, the UAV parameters are compared and analyzed one by one with standard threshold ranges, including:

[0025] A pre-defined standard threshold range is set. The drone parameters are compared with the standard threshold range. If all drone parameters are greater than the maximum value of the standard threshold range, the drone is marked as a first-priority drone and added to the first-priority transmission queue. If all drone parameters are greater than the minimum value of the standard threshold range, or if any drone parameter is greater than the maximum value of the standard threshold range, the drone is marked as a second-priority drone and added to the second-priority transmission queue. If any drone parameter is less than the minimum value of the standard threshold range, the drone is marked as a third-priority drone and added to the third-priority transmission queue.

[0026] To achieve the above objectives, according to another aspect of the present invention, an image transmission device based on an artificial intelligence algorithm is provided, comprising an image acquisition module, an image decomposition module, and an image transmission module;

[0027] The image acquisition module uses remote sensing equipment to acquire raw image data and performs preliminary data cleaning on the images;

[0028] The image decomposition module is used to segment remote sensing images, analyze the regional transmission value and image complexity of remote sensing images, and then obtain the comprehensive image transmission value. The remote sensing images are then labeled based on the comprehensive image transmission value.

[0029] The image transmission module is used to comprehensively evaluate the overall performance of the device and match transmission tasks.

[0030] Compared with the prior art, the beneficial effects of the present invention are:

[0031] 1. This invention employs artificial intelligence algorithms to perform real-time image segmentation and content analysis using a lightweight deep learning model. It identifies important regions in the image and performs differentiated transmission based on the priority of different regions, ensuring that important regions receive higher quality transmission. By monitoring network status in real time and analyzing parameters such as bandwidth, latency, and packet loss rate, the image transmission strategy is dynamically adjusted according to the current network conditions. This includes lossy compression, dynamic adjustment of image compression rate, and prioritizing the transmission of key regions, thereby reducing the impact of network instability on image quality.

[0032] 2. This invention performs real-time performance evaluation of UAVs, dynamically adjusts task allocation based on the comprehensive score of the equipment, prioritizes the allocation of tasks to equipment with better performance, and flexibly adjusts the task execution order according to task priority and network conditions.

[0033] 3. This invention monitors network status, device status, task progress, etc. during the transmission process through a real-time feedback mechanism, and automatically adjusts the transmission strategy to ensure that the transmission task can still be completed efficiently and stably when network and device conditions change. Attached Figure Description

[0034] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an undue limitation of the invention. In the drawings:

[0035] Figure 1 This is a schematic diagram of the method flow of the present invention;

[0036] Figure 2 This is a schematic diagram of the connections between the modules of the present invention. Detailed Implementation

[0037] It should be noted that, unless otherwise specified, the embodiments and features described in the present invention can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0038] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0039] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate for the embodiments of the invention described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0040] To make the objectives and advantages of the present invention clearer, the present invention will be further described below with reference to embodiments; it should be understood that the specific embodiments described herein are merely for explaining the present invention and are not intended to limit the present invention.

[0041] According to embodiments of the present invention, such as Figure 1 As shown, an image transmission method based on artificial intelligence algorithms is provided, which includes the following steps:

[0042] S1: Collect raw remote sensing image data using remote sensing equipment, and record the resolution, spectral information, timestamp, and geographical location of the remote sensing images. Perform preliminary data cleaning on the remote sensing images to ensure the validity and accuracy of the data. Data cleaning includes: using Gaussian filtering to remove image noise and reduce low-frequency interference in the images; and using convolutional neural networks to extract features from the raw images, including edge, texture, and object information, in order to provide basic data for subsequent image processing and classification.

[0043] S2: Segment the remote sensing image, and comprehensively analyze the region optimization transmission value and image complexity of the remote sensing image to obtain a comprehensive transmission value. Label the remote sensing image based on the comprehensive transmission value; the specific steps are as follows:

[0044] Remote sensing images are segmented in real time using deep learning models to obtain several sub-images; the deep learning models can be MobileNetV2, EfficientNet, etc.

[0045] For several sub-images, identify and label the corresponding type regions, where the type regions include cities, roads, farmland, sky, forests, backgrounds, irrigation equipment, etc.; the type region number is denoted as n.

[0046] The location and size of each sub-image in the remote sensing image are determined. The segmented sub-images are then subjected to noise reduction, cropping, normalization, and enhancement. The sub-image number (i = 1, ..., I) is defined, where I is the total number of segments in the remote sensing image. In remote sensing image transmission, different task modes affect the relative importance of different regions. For example, in agricultural monitoring tasks, farmland areas are the core monitoring area and have a higher priority. Therefore, based on the actual needs and application scenarios of the areas, different weight ratios are assigned to different types of regions. Users can freely choose the task mode when performing image transmission. Several preset task modes are available, with r as the task mode number. For example, several preset task modes include post-disaster assessment mode, agricultural monitoring mode, animal observation mode, etc. Different weights are assigned to different types of regions within each preset task mode. The weights corresponding to different types of regions in different task modes are denoted as Rnr. For example, in the post-disaster assessment model, if r is set to 1, then the weight of the urban area is R11, where R11 = 1.5; the weight of the road area is R21, where R21 = 1.2; the weight of the farmland area is R31, where R31 = 0.9; and so on. Specific settings can be customized by those skilled in the art based on actual usage.

[0047] Identify the weight Rnr corresponding to the type region in the preset task mode for each sub-image i, and substitute it into... The regional superiority index Qi of sub-image i is calculated; where Gi is the ratio between the sub-image and the remote sensing image; norm[] represents the normalization function to ensure dimensional consistency; gpi is the high-frequency energy ratio of the sub-image, specifically: after sub-image i is transformed to the frequency domain by Fourier transform (FFT), the energy of the high-frequency components and the total energy are identified, and the ratio of the high-frequency components to the total energy is recorded as the high-frequency energy ratio. It should be noted that: the high-frequency components refer to the parts outside the central region of the sub-image; the clearer the image, the higher the corresponding high-frequency energy ratio. The greater the weight and proportion of a type region, the larger its region-optimal index, facilitating subsequent priority transmission. The region-optimal index Qi of each sub-image i in the remote sensing image is summed to obtain the region-optimal transmission value Qi of the remote sensing image. If the region-optimal transmission value is less than a preset first threshold, the overall transmission value is directly taken as a preset value. If the region-optimal transmission value is greater than or equal to the preset first threshold, the remote sensing image is converted into an 8-bit grayscale image, where each pixel value in the grayscale image ranges from 0 to 255. For example, the number of pixels with a value of 0 (black) in the image is counted, and so on, counting the occurrences of each pixel value in the grayscale image and calculating the probability corresponding to each pixel value, where m is the number of each pixel value, m∈{0,1,2,3,...,255}. The information entropy formula is used... The image complexity HLim is calculated, where L im Let L be the probability of pixel value m; if a pixel value m does not appear, then L...im The probability is zero.

[0048] It should be noted that the preset first threshold value is custom-set by those skilled in the art based on actual use. The image complexity HLim value ranges from 0 to 8; the higher the entropy value, the more dispersed the pixel values ​​in the image, and the higher the image complexity.

[0049] Substitute the values ​​of the localization value (QI) and the image complexity (HLim) of the remote sensing image into the set formula. The overall image transmission value Z is calculated, where λ1 and λ2 are the set weight factors, and norm[] represents the normalization function. As can be seen from the formula, the larger the regional transmission value, the higher the image complexity, and the larger the overall image transmission value, the higher the priority level of transmission.

[0050] A preset image overall transmission threshold range is established, and the image overall transmission value is compared with the set image overall transmission threshold range.

[0051] When the overall image transmission value is greater than the maximum value of the set overall image transmission threshold range, the image is marked as the first priority transmission image;

[0052] When the overall image transmission value is within the set overall image transmission threshold range, the image is marked as the second priority transmission image;

[0053] When the overall image transmission value is less than the minimum value of the set overall image transmission threshold range, the image is marked as the third priority transmission image.

[0054] S3: Evaluate the overall performance of the equipment based on the drone parameters and match the transmission task accordingly; the specific steps are as follows:

[0055] Using several drones as remote sensing image transmission devices, the remaining flight time of all drones is obtained and labeled as Tj, j=1, 2, 3, ..., J, where J is the total number of drones and j is the ID of each drone. The maximum supported bandwidth, real-time available bandwidth, and GPU computing power of each drone are also obtained and labeled as B. 最大 Yj, Bj, and Uj; the current network latency, packet loss rate, and jitter of the drone are measured using network detection tools and labeled as Yj, Pj, and Dj, respectively; it should be noted that jitter refers to the degree of change in data transmission latency in the network.

[0056] Normalize the drone parameters of different dimensions to the [0, 1] interval, including: setting the maximum and minimum flight times of the drone to T. 最大 and T 最小 Using the formula The normalized endurance value Tj1 is calculated; the minimum bandwidth of the drone is set to B. 最小 Using the formula The normalized bandwidth value Bj1 is calculated; the minimum GPU computing power of the drone is set to U. 最小 The maximum GPU computing power of the drone is set to U. 最大 Using the formula The normalized computing power value Uj1 is calculated; the maximum network latency is set to Y. 最大 Set the minimum network latency to Y. 最小 Using the formula The normalized delay value Yj1 is calculated; the maximum packet loss rate is set to P. 最大 Set the minimum packet loss rate to P. 最小 Using the formula The normalized packet loss value Pj1 is calculated; the maximum jitter value is set to D. 最大 Set the minimum value of jitter to D. 最小 Using the formula The normalized jitter value Dj1 is calculated;

[0057] The drone's endurance, bandwidth, computing power, latency, packet loss, and jitter values ​​are removed of their dimensions, and their values ​​are substituted into a predefined linear model to calculate the drone's score Fj. A specific example could be: ; where β1, β2, β3, β4, β5 and β6 are the set weight factors.

[0058] Several task types are defined. When executing a transmission task, the parameter with the highest transmission occupancy is detected. For example, if the transmission task consumes the most bandwidth, it is determined to be a real-time transmission task, and the bandwidth weight β2 is increased to 150%. If the drone consumes the most computing power, it is determined to be a computationally intensive task, and the computing power weight β3 is increased to 150%.

[0059] A standard threshold range is set as [30%, 70%]. The obtained endurance, bandwidth, computing power, latency, packet loss, and jitter values ​​are compared and analyzed one by one with the set standard threshold range. When the drone's endurance, bandwidth, computing power, latency, packet loss, and jitter values ​​are all greater than the maximum value of the set standard threshold range (70%), the drone is marked as a first-priority drone and added to the first-priority transmission queue. When the drone's endurance, bandwidth, computing power, latency, packet loss, and jitter values ​​are all greater than the minimum value of the set standard threshold range (30%), and any one of these values ​​is less than the maximum value of the set standard threshold range (70%), the drone is marked as a second-priority drone and added to the second-priority transmission queue. When any one of the drone's endurance, bandwidth, computing power, latency, packet loss, and jitter values ​​is less than the minimum value of the set standard threshold range (30%), the drone is marked as a third-priority drone and added to the third-priority transmission queue. The criteria for determining the priority transmission queue are as follows: The criteria for determining the second priority transmission queue are: The criteria for determining the third priority transmission queue are: ;

[0060] All priority transmission images are sorted from highest to lowest based on their overall transmission value, and these images are transmitted first. Each transmission queue is sorted in descending order based on the drone's score, and matched sequentially. Drones with higher scores are matched with priority transmission images, such as for real-time disaster monitoring. After a drone accepts a transmission task, it is removed from the transmission queue until the task is completed. Then, the values ​​are recalculated and the drone is added back to the transmission queue. Parameters are re-collected and the queue is refreshed every 5 minutes. If the equipment parameters change abruptly, such as a sudden increase in packet loss rate, the drone is immediately downgraded to the third priority queue. The second priority transmission queue matches second priority transmission images and handles routine tasks, such as periodic inspections of farmland. The third priority transmission queue does not participate in task allocation temporarily and needs maintenance or charging. After maintenance or recharging is completed, parameters are re-collected and the queue is refreshed.

[0061] S4: Utilize AI for real-time monitoring and prediction of network conditions, automatically adjusting image transmission strategies based on network bandwidth, latency, and other factors; specific steps are as follows:

[0062] Set the total network bandwidth for all drones to B. 总 Monitor the current total bandwidth every minute and mark it as B. 当前 If the following conditions are met: If this occurs, the total bandwidth is determined to decrease by 20%; the weight of high-priority areas (such as cities and transportation) is increased proportionally, using the formula... The weight Rnw after the proportional increase is calculated. 新Where E is the set urgency level, E∈{1,2,3}, 1 is the default normal level, 2 is the medium level, and 3 is the highest level;

[0063] The packet loss rate is monitored every 2 minutes. When the packet loss rate exceeds 10%, lossy compression is initiated, using the formula... The adjusted compression ratio C is calculated; if the latency is greater than 80ms or the jitter is greater than 15ms, the compression ratio is further reduced: ;

[0064] If the packet loss rate is greater than 30%, the latency is greater than 100ms, the jitter is greater than 20ms, or the battery life is less than 15 minutes, the device will be immediately moved to the third queue, new tasks will be stopped, and a maintenance alarm will be triggered to notify the operator to inspect or charge the device.

[0065] If a drone in the priority queue malfunctions and the mission is interrupted, the drone with the highest score in the second queue will be automatically assigned to take over. After the mission is migrated, the original mission data packets will be resumed from the point of interruption to avoid duplicate transmission. When the load of the second queue device exceeds 80%, some missions will be migrated to the device with the highest score in the third priority transmission queue.

[0066] The drone score is recalculated and updated every 5 minutes; if the drone parameters change by more than 20%, the score is immediately recalculated; in the priority queue, drones are sorted in descending order of score, and if the scores are the same, the device with longer battery life is selected first; in the second priority queue, drones are sorted in descending order of bandwidth utilization, and if the bandwidth utilization is the same, the device with longer battery life is selected first.

[0067] The task success rate, average latency, and bandwidth utilization are calculated every 5 minutes. If the task success rate is below 90%, the regional priority weight λ1+0.1 is automatically increased. This is adjusted only once every 5 minutes until the next 5 minutes when the calculation is repeated. When the bandwidth utilization is below 70%, the compression ratio C=C-0.05 is reduced. This is adjusted only once every 5 minutes until the next 5 minutes when the calculation is repeated.

[0068] S5: Based on a user interface, it provides real-time monitoring and adjustment to help users efficiently configure, manage, and optimize image transmission tasks; the specific steps are as follows:

[0069] Users select remote sensing images to be transmitted from local devices or the cloud. The browser interface displays a list of images, thumbnails, and basic information, including resolution, shooting time, and type. Users can select images individually or in batches for transmission.

[0070] Users can freely choose the transmission target, such as the target receiving device or the target location, and determine the target by entering the identifier of the receiving device or the location coordinates; it should be noted that the target receiving device includes ground stations, cloud, other drones, etc.

[0071] Set parameters such as image quality (e.g., high quality, medium quality, low quality) and transmission speed (e.g., high-speed transmission, standard transmission, low-speed transmission) according to task requirements; set transmission strategies, such as prioritizing transmission of critical areas, using lossy compression, real-time compression, etc.; display the image transmission status and progress; also display information such as remaining transmission time, current transmission rate, bandwidth utilization, packet loss rate, latency, and percentage of transmission completion; display a transmission progress bar to show the transmission completion status; when network anomalies occur, such as insufficient bandwidth or increased packet loss rate, the user interface automatically displays warnings and generates improvement suggestions, such as adjusting the compression rate and prioritizing transmission areas.

[0072] The system displays real-time changes in network conditions during transmission, including: bandwidth utilization (showing current and historical bandwidth usage); latency (showing real-time changes in transmission latency); packet loss rate (showing real-time monitoring of packet loss rate); and jitter (showing real-time changes in network jitter). If network conditions are poor, such as high latency or high packet loss rate, the user interface generates automatic optimization suggestions or allows users to manually adjust transmission strategies. Furthermore, transmission parameters, such as compression ratio, image quality, and priority, can be adjusted at any time through the user interface to flexibly respond to network changes during transmission.

[0073] This invention employs artificial intelligence algorithms to perform real-time image segmentation and content analysis using a lightweight deep learning model. It identifies important regions in the image and performs differentiated transmission based on the priority of different regions, ensuring that important regions receive higher quality transmission. By monitoring network status in real time and analyzing parameters such as bandwidth, latency, and packet loss rate, the image transmission strategy is dynamically adjusted according to the current network conditions. By using lossy compression, dynamically adjusting the image compression rate, and prioritizing the transmission of key regions, the impact of network instability on image quality is reduced.

[0074] This invention performs real-time performance evaluation on image-based drones, dynamically adjusts task allocation based on the comprehensive scores of the devices, prioritizes tasks for devices with better performance, and flexibly adjusts the task execution order according to task priority and network conditions.

[0075] This invention monitors network status, device status, and task progress during transmission through a real-time feedback mechanism, and automatically adjusts transmission strategies to ensure that transmission tasks can still be completed efficiently and stably when network and device conditions change.

[0076] According to embodiments of the present invention, such as Figure 2As shown, an image transmission device based on artificial intelligence algorithms is also provided. The device includes: an image acquisition module, an image decomposition module, an image transmission module, an intelligent optimization module, and an interactive control module.

[0077] The image acquisition module uses remote sensing equipment to collect raw image data, recording the image resolution, spectral information, timestamps, and geographical location. It also performs preliminary data cleaning to ensure the validity and accuracy of the data. Data cleaning includes: using Gaussian filtering to remove image noise and reduce low-frequency interference in the image; and using convolutional neural networks to extract key features from the image, such as edges, textures, and object information, providing a foundation for subsequent image processing and classification.

[0078] The image decomposition module segments the image and comprehensively analyzes the region priority and image complexity of the image, and allocates transmission tasks accordingly.

[0079] The image transmission module comprehensively evaluates the overall performance of the equipment based on the drone's performance parameters and matches transmission tasks accordingly. Several task types are defined, and during the execution of a transmission task, the module detects the parameter with the highest transmission occupancy. For example, if the transmission task consumes the most bandwidth, it is determined to be a real-time transmission task, and the bandwidth weight β2 is increased to 150%. If the drone's computing power consumption is the highest, it is determined to be a computationally intensive task, and the computing power weight β3 is increased to 150%.

[0080] The intelligent optimization module uses AI to monitor and predict network conditions in real time, and automatically adjusts the image transmission strategy based on network bandwidth, latency and other conditions.

[0081] The interactive control module provides an intuitive and easy-to-use interface, real-time monitoring, and flexible adjustment mechanisms to help users efficiently configure, manage, and optimize image transmission tasks.

[0082] The above are merely embodiments of the present invention and are not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principle of the present invention should be included within the scope of the claims of the present invention.

Claims

1. An image transmission method based on artificial intelligence algorithms, characterized in that, Includes the following steps: Raw remote sensing image data is collected using remote sensing equipment, and preliminary data cleaning is performed on the remote sensing images. The remote sensing image is then segmented, and the regional transmission value and image complexity of the remote sensing image are analyzed to obtain the overall image transmission value, which is then used to label the remote sensing image. The overall performance of the equipment is comprehensively evaluated, and transmission tasks are matched accordingly. The specific process is as follows: All UAV parameters are acquired and normalized to the [0, 1] interval. The normalized UAV parameters are then processed to obtain the UAV's score. Several task types are defined. When executing a transmission task, the parameter with the highest transmission occupancy rate is detected, and it is determined as a real-time transmission task, with its weight increased. The UAV parameters are compared and analyzed one by one with the standard threshold interval to determine the priority transmission queue for the UAVs. The overall image transmission value and the priority transmission queue are matched, and then the corresponding remote sensing image is transmitted.

2. The image transmission method based on artificial intelligence algorithm according to claim 1, characterized in that, The remote sensing image is segmented, and the region optimization values ​​and image complexity of the remote sensing image are analyzed, including: Remote sensing images are segmented in real time using a deep learning model to obtain several sub-images; Identify and label the corresponding type regions for several sub-images; Determine the corresponding weight based on the selected task mode; Based on the determined weights, the area superiority index of each sub-image is calculated. The area superiority indices of each sub-image in the remote sensing image are summed to obtain the area superiority transmission value of the remote sensing image. If the area superiority transmission value is less than the preset first threshold, the comprehensive transmission value is directly taken as the preset value. If the area-optimized transmission value is greater than or equal to the preset first threshold, the remote sensing image will be converted into an 8-bit grayscale image. Identify and count the number of times each pixel value appears in a grayscale image, calculate the probability corresponding to each pixel value, and calculate the image complexity using the information entropy formula. The overall image transmission value of remote sensing images is obtained by analyzing the regional transmission value and image complexity of remote sensing images.

3. The image transmission method based on artificial intelligence algorithm according to claim 2, characterized in that, Labeling the remote sensing image includes: A preset image overall transmission threshold range is established, and the image overall transmission value is compared with the set image overall transmission threshold range. When the overall image transmission value is greater than the maximum value of the set overall image transmission threshold range, the image is marked as the first priority transmission image; When the overall image transmission value is within the set overall image transmission threshold range, the image is marked as the second priority transmission image; When the overall image transmission value is less than the minimum value of the set overall image transmission threshold range, the image is marked as the third priority transmission image.

4. The image transmission method based on artificial intelligence algorithm according to claim 1, characterized in that, The normalized UAV parameters are processed to obtain the UAV's score, including: The drone's endurance, bandwidth, computing power, latency, packet loss, and jitter values ​​are input into a linear model to obtain the drone's score.

5. The image transmission method based on artificial intelligence algorithm according to claim 1, characterized in that, The parameters of the drone are compared and analyzed one by one with the standard threshold range, including: A pre-defined standard threshold range is set. The drone parameters are compared with the standard threshold range. If all drone parameters are greater than the maximum value of the standard threshold range, the drone is marked as a first-priority drone and added to the first-priority transmission queue. If all drone parameters are greater than the minimum value of the standard threshold range, or if any drone parameter is greater than the maximum value of the standard threshold range, the drone is marked as a second-priority drone and added to the second-priority transmission queue. If any drone parameter is less than the minimum value of the standard threshold range, the drone is marked as a third-priority drone and added to the third-priority transmission queue.

6. An image transmission device based on an artificial intelligence algorithm, characterized in that... An image transmission method based on an artificial intelligence algorithm according to any one of claims 1-5 includes an image acquisition module, an image decomposition module, and an image transmission module; The image acquisition module uses remote sensing equipment to acquire raw image data and performs preliminary data cleaning on the images; The image decomposition module is used to segment remote sensing images, analyze the regional transmission value and image complexity of remote sensing images, and then obtain the comprehensive image transmission value. The remote sensing images are then labeled based on the comprehensive image transmission value. The image transmission module is used to comprehensively evaluate the overall performance of the device and match transmission tasks.

Images