Image transmission method and system based on priority sorting

Abstract

The present invention discloses an image transmission method and system based on priority sorting, the method comprising: S1: classifying images based on a convolutional neural network; S2: weighting the attention of the pixels in the image and calculating the image center of gravity position; S3: calculating the total time cost of the image based on the compression, storage, decompression and transmission time of the image; S4: combining the transmission cost based on the image category, the transmission cost based on the image center of gravity position and the total time cost of the image, calculating the total image transmission cost, and calculating the optimal image transmission cost; S5: determining the transmission priority of images adjacent in time sequence based on the optimal image transmission cost, and recording them using a graph structure; S6: using the graph structure to transmit the image. The present invention constructs a reasonable total image transmission cost, and optimizes the image transmission priority by solving the objective function; using the graph structure, the image transmission priority is recorded and read, and finally the image is efficiently transmitted.

Description

Technical Field

[0001] The present invention relates to the technical field of image transmission, and in particular to an image transmission method and system based on priority sorting. Background Art

[0002] In the image transmission scenario, images are transmitted after being sorted according to image timing, image content, image importance or target location. Sorting and transmitting images is conducive to giving priority to the transmission of images containing key content, important images or images containing targets, and maximizing the benefits of image transmission under limited bandwidth and storage space; it is also conducive to the reasonable allocation of computing resources and network bandwidth to ensure the smooth execution of key tasks; and maintaining stable image transmission performance in the case of bandwidth fluctuations or network congestion. At present, the common solutions for image sorting in image transmission scenarios are: time-based sorting transmission and image processing-based sorting transmission. Among them, time-based sorting transmission mainly sorts and transmits images according to the timing of images. Image processing-based sorting transmission mainly extracts content features, regional features, and target features in images, and sorts and transmits images based on features.

[0003] Among the above image transmission methods, the purely time-based sorting transmission is suitable for real-time image transmission scenarios, but there are also cases of data backlog, long queue waiting time, and transmission delay. The sorting transmission based on image processing may cause large changes in the image order, destroy the data correlation between image sequences, and cause difficulties in image processing and sorting at the receiving end, increasing the risk of erroneous transmission. Summary of the invention

[0004] In view of this, the present invention, based on the existing image transmission order, comprehensively considers image category, image importance, and total time cost of images, and optimizes the transmission order of temporally adjacent images; on the premise of ensuring the overall stability of image transmission, fine-tunes the local order of image transmission to improve image transmission efficiency and quality.

[0005] The technical solution proposed by the present invention is an image transmission method based on priority sorting, comprising the following steps:

[0006] S1: Classify the image based on the convolutional neural network to obtain the image category;

[0007] S2: Obtain the attention weight of the pixel in the image based on the reference pixel, then perform attention weighting on the pixel in the image and calculate the center of gravity of the image;

[0008] S3: Combined with the image compression ratio, the image compression time, storage time, decompression time and transmission time are calculated respectively to obtain the total time cost of the image;

[0009] S4: Calculate the transmission cost based on the image center of gravity position in combination with the image compression ratio; Calculate the total image transmission cost in combination with the transmission cost based on the image category, the transmission cost based on the image center of gravity position and the total time cost of the image, and obtain the optimal image compression ratio based on the principle of minimizing the total image transmission cost. The total image transmission cost at this time is the optimal image transmission cost;

[0010] S5: Determine the transmission priority of temporally adjacent images based on the optimal image transmission cost and record it in a graph structure, including:

[0011] For images M and N that are adjacent in time sequence, when the optimal image transmission cost of image M is lower than the optimal image transmission cost of image N, the transmission priority of image M is higher than that of image N; when the optimal image transmission cost of image M is higher than the optimal image transmission cost of image N, the transmission priority of image N is higher than that of image M; when the optimal image transmission cost of image M is equal to the optimal image transmission cost of image N, the transmission priority of image M is equal to that of image N;

[0012] A graph structure is used to record the sequence numbers of all images and the transmission priority of images that are adjacent in time sequence. Each node in the graph structure represents an image, and the node set is expressed as:

[0013] ;

[0014] in, Represents a collection of nodes. Represents a node, Represents the number of elements in the node set, is equal to the number of images to be transmitted;

[0015] Establish edges between adjacent nodes and the weight of the edges According to the node The corresponding image e and node The transmission priority between the corresponding images f is determined by: , are two adjacent nodes in the node set; image e and image f are temporally adjacent, and e and f represent the sequence numbers of the images;

[0016] ;

[0017] S6: Utilize the graph structure to transmit images based on transmission priority.

[0018] Preferably, the step S1 comprises:

[0019] Construct a convolutional neural network and train it. When the network weights converge, the convolutional neural network training is completed and a convolutional neural network model is obtained.

[0020] The image is input into the convolutional neural network model and the image category is output.

[0021] Preferably, the step S2 comprises:

[0022] S21: Calculate the pixels in the image The gradient is:

[0023] ;

[0024] in, Represents the pixel points in the image The gradient of Represents the pixel points in the image The amplitude, Represents the pixel points in the image The magnitude is Directional derivative, Represents the pixel points in the image The magnitude is Partial derivative of direction;

[0025] Determine the reference pixel for calculating the attention weight and calculate the pixel in the image Similarity with reference pixels:

[0026] ;

[0027] in, Represents the pixel points in the image and reference pixel The similarity of Indicates the reference pixel The gradient of Indicates the absolute value.

[0028] Pixels in the image Relative to the reference pixel The attention weight is expressed as:

[0029] ;

[0030] in, Represents the pixel points in the image Relative to the reference pixel The attention weight, Represents the pixel points in the image Sum the similarities with all reference pixels;

[0031] S22: Perform attention weighting on the pixels in the image to obtain:

[0032] ;

[0033] in, Represents the pixels in the image The result obtained after attention weighting is: Indicates the reference pixel the magnitude of

[0034] The center of gravity of the image is:

[0035] ;

[0036] in, Indicates the center of gravity of the image.

[0037] Preferably, step S3 comprises:

[0038] The compression time of the image is:

[0039] ;

[0040] in, Indicates the compression time of the image, Indicates the image compression ratio, is the compression time coefficient;

[0041] The image storage time is:

[0042] ;

[0043] in, Indicates the storage time of the image. is the storage time coefficient;

[0044] The decompression time of the image is:

[0045] ;

[0046] in, Indicates the decompression time of the image, is the decompression time coefficient;

[0047] The image transmission time is:

[0048] ;

[0049] in, Indicates the transmission time of the image, is the transmission time coefficient;

[0050] The total time cost for the image is:

[0051] ;

[0052] in, Represents the total time cost of the image.

[0053] Preferably, step S4 comprises:

[0054] S41: The transmission cost based on the image center of gravity position is expressed as:

[0055] ;

[0056] in, represents the transmission cost based on the image center position, represents the expected center of gravity position of the image, represents the image center position obtained in step S2,

[0057] represents the difference between the center of gravity position of the image obtained in step S2 and the expected center of gravity position of the image;

[0058] The total transmission cost of the image is:

[0059] ;

[0060] in, represents the total transmission cost of the image, represents the total time cost coefficient of the image, represents the total time cost of the image obtained in step S3, represents the transmission cost coefficient based on the image center position, represents the transmission cost coefficient based on image category, represents the transmission cost based on image category, The value of is determined based on the image category obtained in step S1;

[0061] S42: Based on the principle of minimizing the total image transmission cost, construct the objective function and constraints:

[0062] ;

[0063] Among them, the objective function By minimizing the total transmission cost To optimize image compression ratio , Indicates the image compression ratio The lower limit of Indicates the image compression ratio The upper limit of

[0064] Solve to obtain the optimal image compression ratio , the total image transmission cost at this time is the optimal image transmission cost.

[0065] Preferably, step S6 comprises:

[0066] S61: Mark all nodes as not transmitted, and let the first node For the current node ;

[0067] S62: Check the current node Neighbor nodes exists;

[0068] S63: If it does not exist, transmit the current node The corresponding image is generated, and the current node is marked as transmitted, and the operation ends;

[0069] If it exists, determine the current node With neighbor nodes Do the edges between them satisfy ;

[0070] S64: If not satisfied, transfer the current node The corresponding image is obtained, and the current node is marked as transmitted, and the process goes to step S66;

[0071] If satisfied, determine the neighbor node Whether it is marked as not transmitted;

[0072] S65: If the neighbor node Mark as not transmitted, and jump to step S66;

[0073] If the neighbor node Mark as transferred, transfer the current node The corresponding image is obtained, and the current node is marked as transmitted, and the process goes to step S66;

[0074] S66: Select neighbor nodes For the current node , jump to step S62.

[0075] The present invention also provides an image transmission system based on priority sorting, comprising:

[0076] Image classification module: build a convolutional neural network, train the convolutional neural network, obtain a convolutional neural network model, and the convolutional neural network outputs image categories;

[0077] Image center of gravity module: determine the reference pixel, calculate the similarity between the pixel in the image and the reference pixel, calculate the attention weight of the pixel in the image relative to the reference pixel, perform attention weighting on the pixel in the image, and calculate the image center of gravity based on the attention weighting;

[0078] Image total time cost module: calculates the image compression time, image storage time, image decompression time, image transmission time, and image total time cost;

[0079] Image transmission cost module: calculates the total transmission cost of the image, obtains the optimal image compression ratio based on the principle of minimizing the total transmission cost of the image, and obtains the optimal image transmission cost;

[0080] Transmission priority module: compares the optimal image transmission costs of temporally adjacent images, and determines the transmission priority of temporally adjacent images based on the comparison results; uses a graph structure to record the sequence numbers of all images and the transmission priority of temporally adjacent images;

[0081] Image transmission module: mark all nodes as not transmitted, determine whether the neighbor node of the current node exists, determine the edge between the current node and the neighbor node, determine whether the neighbor node is marked as not transmitted, transmit the image corresponding to the current node, mark the current node as transmitted, and select the neighbor node as the current node.

[0082] Beneficial effects:

[0083] The present invention obtains a more reasonable total image transmission cost based on comprehensive consideration of image category, image importance, and total image time cost; constructs an objective function and optimizes and solves to obtain the optimal image compression ratio, which is conducive to setting pictures according to the optimal image compression ratio, reducing the total image transmission cost, and alleviating the load; the present invention uses the optimal image transmission cost as a measurement standard to determine the transmission priority of adjacent images, which is conducive to optimizing the local image transmission order, improving the transmission efficiency in the short term, and ensuring the overall stability of image transmission; the present invention uses a graph structure, which is conducive to sequentially recording and reading image sequence numbers and the transmission priority of sequentially adjacent images.

[0084] The present invention only compares the optimal image transmission costs of temporally adjacent images, which reduces the workload and improves the processing efficiency. Based on the comparison, it is simpler and more convenient to record the transmission order of temporally adjacent images. During the image transmission in step S6, the image corresponding to the node with the highest local transmission order priority is obtained. The local optimal method balances the computing resources and the optimal solution to the problem, and is a feasible solution.

[0085] The present invention calculates the attention weights of pixels in an image based on reference pixels, which is helpful for measuring the importance of pixels in the image; attention weighting is performed on pixels in the image to obtain the center of gravity position of the image, which is helpful for measuring the importance of the image according to the center of gravity position of the image. BRIEF DESCRIPTION OF THE DRAWINGS

[0086] Figure 1 A schematic flow chart of an image transmission method based on priority sorting provided by an embodiment of the present invention.

[0087] Figure 2 A schematic diagram of a process for sequentially selecting images with the highest priority for transmission using a graph structure provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0088] The present invention is further described below in conjunction with the accompanying drawings, but the present invention is not limited in any way. Any changes or substitutions made based on the teachings of the present invention belong to the protection scope of the present invention.

[0089] Embodiment 1:

[0090] A method for image transmission based on priority sorting, such as Figure 1 As shown, the following steps are included:

[0091] S1: Classify the image based on the convolutional neural network to obtain the image category:

[0092] Construct a convolutional neural network and train it. When the network weights converge, the convolutional neural network training is completed and a convolutional neural network model is obtained.

[0093] The image is input into the convolutional neural network model and the image category is output.

[0094] S2: Get the attention weight of the pixel in the image based on the reference pixel, then perform attention weighting on the pixel in the image and calculate the center of gravity of the image:

[0095] S21: Calculate the pixels in the image The gradient is:

[0096] ;

[0097] in, Represents the pixel points in the image The gradient of Represents the pixel points in the image The amplitude, Represents the pixel points in the image The magnitude is Directional derivative, Represents the pixel points in the image The magnitude is Partial derivative of direction;

[0098] Determine the reference pixel for calculating the attention weight and calculate the pixel in the image Similarity with reference pixels:

[0099] ;

[0100] in, Represents the pixel points in the image and reference pixel The similarity of Indicates the reference pixel The gradient of Indicates the absolute value.

[0101] Pixels in the image Relative to the reference pixel The attention weight is expressed as:

[0102] ;

[0103] in, Represents the pixel points in the image Relative to the reference pixel The attention weight, Represents the pixel points in the image Sum the similarities with all reference pixels;

[0104] S22: Perform attention weighting on the pixels in the image to obtain:

[0105] ;

[0106] in, Represents the pixels in the image The result obtained after attention weighting is: Indicates the reference pixel the magnitude of

[0107] The center of gravity of the image is:

[0108] ;

[0109] in, Indicates the center of gravity of the image.

[0110] In the embodiment of the present invention, the reference pixel point is generally selected from the pixel point at the center of the background image, or the pixel point at the edge of the target image, or directly specified; the reference pixel point with significant features is selected. , calculate the pixel points in the image Relative to the reference pixel The similarity of the pixels in the image Relative to the reference pixel The attention weight is the pixel With reference pixel The normalized similarity of

[0111] In the embodiment of the present invention, the pixel point in the image and reference pixel The closer the gradients are, the closer the similarity is. The closer to 1, the pixel points in the image and reference pixel The larger the gradient difference, the greater the similarity The closer to 0.

[0112] S3: Combined with the image compression ratio, the image compression time, storage time, decompression time and transmission time are calculated respectively to obtain the total time cost of the image:

[0113] The compression time of the image is:

[0114] ;

[0115] in, Indicates the compression time of the image, Indicates the image compression ratio, is the compression time coefficient;

[0116] The image storage time is:

[0117] ;

[0118] in, Indicates the storage time of the image. is the storage time coefficient;

[0119] The decompression time of the image is:

[0120] ;

[0121] in, Indicates the decompression time of the image, is the decompression time coefficient;

[0122] The image transmission time is:

[0123] ;

[0124] in, Indicates the transmission time of the image, is the transmission time coefficient;

[0125] The total time cost for the image is:

[0126] ;

[0127] in, represents the total time cost of the image;

[0128] In the embodiment of the present invention, the higher the compression ratio, the longer the image compression time, the shorter the image storage time, the longer the image decompression time, the shorter the image transmission time, and the total time cost of the image does not necessarily change.

[0129] S4: Calculate the transmission cost based on the image center of gravity position in combination with the image compression ratio; calculate the total image transmission cost in combination with the transmission cost based on the image category, the transmission cost based on the image center of gravity position, and the total time cost of the image, and obtain the optimal image compression ratio based on the principle of minimizing the total image transmission cost. The total image transmission cost at this time is the optimal image transmission cost:

[0130] S41: The transmission cost based on the image center of gravity position is expressed as:

[0131] ;

[0132] in, represents the transmission cost based on the image center position, represents the expected center of gravity position of the image, represents the image center position obtained in step S2,

[0133] represents the difference between the center of gravity position of the image obtained in step S2 and the expected center of gravity position of the image;

[0134] The total transmission cost of the image is:

[0135] ;

[0136] in, represents the total transmission cost of the image, represents the total time cost coefficient of the image, represents the total time cost of the image obtained in step S3, represents the transmission cost coefficient based on the image center position, represents the transmission cost coefficient based on image category, represents the transmission cost based on image category, The value of is determined based on the image category obtained in step S1;

[0137] S42: Based on the principle of minimizing the total image transmission cost, construct the objective function and constraints:

[0138] ;

[0139] Among them, the objective function By minimizing the total transmission cost To optimize image compression ratio , Indicates the image compression ratio The lower limit of Indicates the image compression ratio The upper limit of

[0140] Solve to obtain the optimal image compression ratio , the total image transmission cost at this time is the optimal image transmission cost;

[0141] In the embodiment of the present invention, the expected center of gravity position of the image containing the target is at the location of the target; the larger the difference between the center of gravity position of the image obtained in step S2 and the expected center of gravity position of the image, the less accurate the target positioning is, and therefore the transmission cost based on the center of gravity position of the image is The bigger;

[0142] In the embodiment of the present invention, in step S41 The value of depends on the image category; when the image category is background, ; The more important the image category is, The larger the value, The specific value of is specified by the user according to the actual situation.

[0143] S5: Determine the transmission priority of temporally adjacent images based on the optimal image transmission cost and record it in a graph structure, including:

[0144] For images M and N that are adjacent in time sequence, when the optimal image transmission cost of image M is lower than the optimal image transmission cost of image N, the transmission priority of image M is higher than that of image N; when the optimal image transmission cost of image M is higher than the optimal image transmission cost of image N, the transmission priority of image N is higher than that of image M; when the optimal image transmission cost of image M is equal to the optimal image transmission cost of image N, the transmission priority of image M is equal to that of image N;

[0145] A graph structure is used to record the sequence numbers of all images and the transmission priority of images that are adjacent in time sequence. Each node in the graph structure represents an image, and the node set is expressed as:

[0146] ;

[0147] in, Represents a collection of nodes. Represents a node, Represents the number of elements in the node set, is equal to the number of images to be transmitted;

[0148] Establish edges between adjacent nodes and the weight of the edges According to the node The corresponding image e and node The transmission priority between the corresponding images f is determined by: , are two adjacent nodes in the node set; image e and image f are temporally adjacent, and e and f represent the sequence numbers of the images;

[0149] ;

[0150] S6: Using the graph structure, images are transmitted based on transmission priority:

[0151] S61: Mark all nodes as not transmitted, and let the first node For the current node ;

[0152] S62: Check the current node Neighbor nodes Does it exist?

[0153] S63: If it does not exist, transmit the current node The corresponding image is generated, and the current node is marked as transmitted, and the operation ends;

[0154] If it exists, determine the current node With neighbor nodes Do the edges between them satisfy ;

[0155] S64: If not satisfied, transfer the current node The corresponding image is obtained, and the current node is marked as transmitted, and the process goes to step S66;

[0156] If satisfied, determine the neighbor node Whether it is marked as not transmitted;

[0157] S65: If the neighbor node Mark as not transmitted, and jump to step S66;

[0158] If the neighbor node Mark as transferred, transfer the current node The corresponding image is obtained, and the current node is marked as transmitted, and the process goes to step S66;

[0159] S66: Select neighbor nodes For the current node , jump to step S62.

[0160] Embodiment 2: The present invention also provides an image transmission system based on priority sorting, comprising the following six modules:

[0161] Image classification module: build a convolutional neural network, train the convolutional neural network, obtain a convolutional neural network model, and the convolutional neural network outputs image categories;

[0162] Image center of gravity module: determine the reference pixel, calculate the similarity between the pixel in the image and the reference pixel, calculate the attention weight of the pixel in the image relative to the reference pixel, perform attention weighting on the pixel in the image, and calculate the image center of gravity based on the attention weighting;

[0163] Image total time cost module: calculates the image compression time, image storage time, image decompression time, image transmission time, and image total time cost;

[0164] Image transmission cost module: calculates the total transmission cost of the image, obtains the optimal image compression ratio based on the principle of minimizing the total transmission cost of the image, and obtains the optimal image transmission cost;

[0165] Transmission priority module: compares the optimal image transmission costs of temporally adjacent images, and determines the transmission priority of temporally adjacent images based on the comparison results; uses a graph structure to record the sequence numbers of all images and the transmission priority of temporally adjacent images;

[0166] Image transmission module: mark all nodes as not transmitted, determine whether the neighbor node of the current node exists, determine the edge between the current node and the neighbor node, determine whether the neighbor node is marked as not transmitted, transmit the image corresponding to the current node, mark the current node as transmitted, and select the neighbor node as the current node.

[0167] It should be noted that the serial numbers of the above embodiments of the present invention are only for description and do not represent the advantages and disadvantages of the embodiments. And the terms "including", "comprising" or any other variants thereof in this article are intended to cover non-exclusive inclusion, so that a process, device, article or method including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, device, article or method. In the absence of further restrictions, an element defined by the sentence "including a ..." does not exclude the presence of other identical elements in the process, device, article or method including the element.

[0168] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) as described above, and includes a number of instructions for a terminal device (which can be a mobile phone, a computer, a server, or a network device, etc.) to execute the methods described in each embodiment of the present invention.

[0169] The above are only preferred embodiments of the present invention, and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A method for image transmission based on priority sorting, characterized in that: The method comprises: S1: Classify the image based on the convolutional neural network to obtain the image category; S2: Obtain the attention weight of the pixel in the image based on the reference pixel, then perform attention weighting on the pixel in the image and calculate the center of gravity of the image; S3: Combined with the image compression ratio, the image compression time, storage time, decompression time and transmission time are calculated respectively to obtain the total time cost of the image; S4: Calculate the transmission cost based on the image center of gravity position in combination with the image compression ratio; Calculate the total image transmission cost in combination with the transmission cost based on the image category, the transmission cost based on the image center of gravity position and the total time cost of the image, and obtain the optimal image compression ratio based on the principle of minimizing the total image transmission cost. The total image transmission cost at this time is the optimal image transmission cost; S5: Determine the transmission priority of temporally adjacent images based on the optimal image transmission cost and record it in a graph structure, including: For images M and N that are adjacent in time sequence, when the optimal image transmission cost of image M is lower than the optimal image transmission cost of image N, the transmission priority of image M is higher than that of image N; when the optimal image transmission cost of image M is higher than the optimal image transmission cost of image N, the transmission priority of image N is higher than that of image M; when the optimal image transmission cost of image M is equal to the optimal image transmission cost of image N, the transmission priority of image M is equal to that of image N; A graph structure is used to record the sequence numbers of all images and the transmission priority of images that are adjacent in time sequence. Each node in the graph structure represents an image, and the node set is expressed as: ； in, Represents a collection of nodes. Represents a node, Represents the number of elements in the node set, is equal to the number of images to be transmitted; Establish edges between adjacent nodes and the weight of the edges According to the node The corresponding image e and node The transmission priority between the corresponding images f is determined by: , are two adjacent nodes in the node set; image e and image f are temporally adjacent, and e and f represent the sequence numbers of the images; ； S6: Using the graph structure, images are transmitted based on transmission priority, including: Mark all nodes as not transmitted, determine whether the neighbor node of the current node exists, determine the edge between the current node and the neighbor node, determine whether the neighbor node is marked as not transmitted, transmit the image corresponding to the current node, mark the current node as transmitted, and select the neighbor node as the current node.

2. The image transmission method based on priority sorting according to claim 1, characterized in that: The step S1 comprises: Construct a convolutional neural network and train it. When the network weights converge, the convolutional neural network training is completed and a convolutional neural network model is obtained. The image is input into the convolutional neural network model and the image category is output.

3. The image transmission method based on priority sorting according to claim 1, characterized in that: The step S2 comprises: S21: Calculate the pixels in the image The gradient is: ； in, Represents the pixel points in the image The gradient of Represents the pixel points in the image The amplitude, Represents the pixel points in the image The magnitude is Directional derivative, Represents the pixel points in the image The magnitude is Partial derivative of direction; Determine the reference pixel for calculating the attention weight and calculate the pixel in the image Similarity with reference pixels: ； in, Represents the pixel points in the image and reference pixel The similarity of Indicates the reference pixel The gradient of Indicates the absolute value. Pixels in the image Relative to the reference pixel The attention weight is expressed as: ； in, Represents the pixel points in the image Relative to the reference pixel The attention weight, Represents the pixel points in the image Sum the similarities with all reference pixels; S22: Perform attention weighting on the pixels in the image to obtain: ； in, Represents the pixels in the image The result obtained after attention weighting is: Indicates the reference pixel the magnitude of The center of gravity of the image is: ； in, Indicates the center of gravity of the image.

4. The image transmission method based on priority sorting according to claim 1, characterized in that: The step S3 comprises: The compression time of the image is: ； in, Indicates the compression time of the image, Indicates the image compression ratio, is the compression time coefficient; The image storage time is: ； in, Indicates the storage time of the image. is the storage time coefficient; The decompression time of the image is: ； in, Indicates the decompression time of the image, is the decompression time coefficient; The image transmission time is: ； in, Indicates the transmission time of the image, is the transmission time coefficient; The total time cost for the image is: ； in, Represents the total time cost of the image.

5. The image transmission method based on priority sorting according to claim 1, characterized in that: The step S4 comprises: S41: The transmission cost based on the image center of gravity position is expressed as: ； in, represents the transmission cost based on the image center position, represents the expected center of gravity position of the image, represents the image center position obtained in step S2, represents the difference between the center of gravity position of the image obtained in step S2 and the expected center of gravity position of the image; The total transmission cost of the image is: ； in, represents the total transmission cost of the image, represents the total time cost coefficient of the image, represents the total time cost of the image obtained in step S3, represents the transmission cost coefficient based on the image center position, represents the transmission cost coefficient based on image category, represents the transmission cost based on image category, The value of is determined based on the image category obtained in step S1; S42: Based on the principle of minimizing the total image transmission cost, construct the objective function and constraints: ； Among them, the objective function By minimizing the total transmission cost To optimize image compression ratio , Indicates the image compression ratio The lower limit of Indicates the image compression ratio The upper limit of Solve to obtain the optimal image compression ratio , the total image transmission cost at this time is the optimal image transmission cost.

6. The image transmission method based on priority sorting according to claim 1, characterized in that: The step S6 comprises: S61: Mark all nodes as not transmitted, and let the first node For the current node ; S62: Check the current node Neighbor nodes exists; S63: If it does not exist, transfer the current node The corresponding image is generated, and the current node is marked as transmitted, and the operation ends; If it exists, determine the current node With neighbor nodes Do the edges between them satisfy ; S64: If not satisfied, transfer the current node The corresponding image is obtained, and the current node is marked as transmitted, and the process goes to step S66; If satisfied, determine the neighbor node Whether it is marked as not transmitted; S65: If the neighbor node Mark as not transmitted, and jump to step S66; If the neighbor node Mark as transferred, transfer the current node The corresponding image is obtained, and the current node is marked as transmitted, and the process goes to step S66; S66: Select neighbor nodes For the current node , jump to step S62.

7. An image transmission system based on priority sorting, characterized in that: include: Image classification module: build a convolutional neural network, train the convolutional neural network, obtain a convolutional neural network model, and the convolutional neural network outputs image categories; Image center of gravity module: determine the reference pixel, calculate the similarity between the pixel in the image and the reference pixel, calculate the attention weight of the pixel in the image relative to the reference pixel, perform attention weighting on the pixel in the image, and calculate the image center of gravity based on the attention weighting; Image total time cost module: calculates the image compression time, image storage time, image decompression time, image transmission time, and image total time cost; Image transmission cost module: calculates the total transmission cost of the image, obtains the optimal image compression ratio based on the principle of minimizing the total transmission cost of the image, and obtains the optimal image transmission cost; Transmission priority module: compares the optimal image transmission costs of temporally adjacent images, and determines the transmission priority of temporally adjacent images based on the comparison results; uses a graph structure to record the sequence numbers of all images and the transmission priority of temporally adjacent images; Image transmission module: mark all nodes as not transmitted, determine whether the neighbor node of the current node exists, determine the edge between the current node and the neighbor node, determine whether the neighbor node is marked as not transmitted, transmit the image corresponding to the current node, mark the current node as transmitted, and select the neighbor node as the current node; To implement an image transmission method based on priority sorting as described in any one of claims 1-6.

Images