Automated cargo sorting method for port cargo loading and unloading

Abstract

The present application provides an automated cargo sorting method for port cargo loading and unloading, and relates to the field of cargo sorting technology. The method includes: obtaining the appearance characteristics of the loaded and unloaded cargo through the port cargo management system, and extracting and labeling multidimensional features with visual sensing equipment. According to the distribution of target ports, a port path map model is constructed, and the sorting path is planned based on this to achieve automated sorting. This application can solve the technical problems of the prior art that the low degree of automation in the sorting process, low recognition accuracy, and unreasonable sorting path planning lead to low efficiency and increased error rate in port cargo sorting, which further affects the overall operating efficiency and logistics costs of the port. The sorting efficiency and accuracy are greatly improved, the sorting path is optimized, and the overall performance of port logistics is improved.

Description

Technical Field

[0001] The present application relates to the technical field of cargo sorting, and in particular to an automated cargo sorting method for port cargo loading and unloading. Background Art

[0002] With the continuous development of global trade, the cargo throughput of ports is increasing day by day. The traditional cargo sorting method mainly relies on manual labor, which is not only inefficient but also prone to errors.

[0003] At present, most of the existing port cargo sorting methods still rely on traditional manual operations or semi-automatic equipment, which have obvious limitations when handling large quantities of diverse cargo. First, manual sorting is inefficient and cannot meet the needs of efficient operation of modern ports. Secondly, misidentification and misclassification are prone to occur during manual sorting, which increases the error rate of cargo handling. Furthermore, existing sorting systems often lack intelligent path planning and optimization functions, resulting in unreasonable sorting paths, further reducing sorting efficiency.

[0004] To sum up, the existing technology has low automation level in the sorting process, low recognition accuracy and unreasonable sorting path planning, which leads to low efficiency and increased error rate in port cargo sorting, further affecting the overall operational efficiency and logistics costs of the port.

[0005] By introducing an automated and intelligent sorting system, the sorting efficiency and accuracy can be greatly improved, and the sorting path can be optimized, thereby reducing operating costs and improving the overall performance of port logistics. Summary of the invention

[0006] The purpose of this application is to provide an automated cargo sorting method for port cargo loading and unloading, in order to solve the technical problems in the prior art that the port cargo sorting efficiency is low, the error rate is increased, and the overall port operation efficiency and logistics costs are further affected due to the low degree of automation in the sorting process, low recognition accuracy, and unreasonable sorting path planning.

[0007] In view of the above problems, the present application provides an automated cargo sorting method for port cargo loading and unloading.

[0008] The present application provides a method for automated cargo sorting for port cargo loading and unloading, wherein the method comprises: acquiring target cargo for loading and unloading through a port cargo management system, and obtaining a preset cargo convolution feature set according to an appearance application standard of the target cargo for loading and unloading; acquiring collected image information of the target cargo for loading and unloading using a visual sensing device, performing multi-dimensional feature extraction on the collected image information based on the preset cargo convolution feature set, and obtaining a cargo appearance feature set; constructing a cargo sorting label library, and labeling the cargo appearance feature set based on the cargo sorting label library to obtain cargo sorting attribute label information; acquiring target port distribution information, performing graph model abstraction on the target port distribution information, and generating a port path graph model; performing sorting path planning on the cargo sorting attribute label information based on the port path graph model, and obtaining a target cargo sorting path, and automatically sorting the target cargo for loading and unloading through the target cargo sorting path.

[0009] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0010] The target cargo to be loaded and unloaded is obtained through the port cargo management system, and a preset cargo convolution feature set is obtained according to the appearance application standard of the target cargo to be loaded and unloaded; the collected image information of the target cargo to be loaded and unloaded is obtained by using visual sensing equipment, and multi-dimensional feature extraction is performed on the collected image information based on the preset cargo convolution feature set to obtain a cargo appearance feature set; a cargo sorting label library is constructed, and the cargo appearance feature set is labeled based on the cargo sorting label library to obtain cargo sorting attribute label information; the target port distribution information is obtained, and the target port distribution information is abstracted by a graph model to generate a port path graph model; the cargo sorting attribute label information is sorted based on the port path graph model to plan the sorting path, and the target cargo sorting path is obtained, and the target cargo to be loaded and unloaded is automatically sorted through the target cargo sorting path, which effectively solves the technical problems of the prior art that the low degree of automation of the sorting process, the low recognition accuracy and the unreasonable sorting path planning lead to low port cargo sorting efficiency and increased error rate, which further affects the overall operation efficiency and logistics cost of the port. The sorting efficiency and accuracy are greatly improved, the sorting path is optimized, and the overall performance of port logistics is improved.

[0011] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented according to the contents of the specification, and in order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are specifically cited below. It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present application, nor is it intended to limit the scope of the present application. Other features of the present application will become easy to understand through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] In order to more clearly illustrate the technical solutions in the present application or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are only exemplary, and for ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying any creative work.

[0013] Figure 1 This is a flow chart of the automated cargo sorting method for port cargo loading and unloading used in this application;

[0014] Figure 2 This is a schematic diagram of the process of determining a set of appearance features of goods in the automated cargo sorting method for port cargo loading and unloading in this application. DETAILED DESCRIPTION

[0015] This application provides an automated cargo sorting method for port cargo loading and unloading, which solves the technical problems of low efficiency and increased error rate in port cargo sorting due to low automation of the sorting process, low recognition accuracy, and unreasonable sorting path planning in the prior art, further affecting the overall operational efficiency and logistics costs of the port. It greatly improves sorting efficiency and accuracy, optimizes sorting paths, and improves the overall performance of port logistics.

[0016] Below, the technical solutions in the present application will be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all of the embodiments of the present application. It should be understood that the present application is not limited to the example embodiments described herein. Based on the embodiments of the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present application. It should also be noted that, for the convenience of description, only the parts related to the present application are shown in the accompanying drawings, rather than all of them. Embodiment 1

[0017] Please refer to the attached Figure 1 The present application provides a method for automated cargo sorting for port cargo loading and unloading, wherein the method specifically comprises the following steps:

[0018] S1: Obtain target loading and unloading cargo through the port cargo management system, and obtain a preset cargo convolution feature set according to the appearance application standard of the target loading and unloading cargo.

[0019] Specifically, sample images of various types of goods are collected, including different angles, lighting conditions, and cargo states. According to the characteristics and classification requirements of the goods, appearance application standards are formulated, including the shape, color, texture, label position, etc. of the goods. The collected cargo images are preprocessed, including image scaling, cropping, denoising, etc. A preset cargo convolution feature set is obtained, which includes the standard color, size, and shape convolution features of the target cargo to be loaded and unloaded.

[0020] S2: using visual sensing equipment to obtain the collected image information of the target loading and unloading cargo, performing multi-dimensional feature extraction on the collected image information based on the preset cargo convolution feature set, and obtaining a cargo appearance feature set.

[0021] Specifically, when the target loading and unloading cargo enters the camera's field of view, the visual sensing device captures the cargo's image information in real time. The acquired images are preprocessed as necessary, such as denoising, enhancement, and standardization, to improve the image quality and prepare for feature extraction. Using deep learning technology, it is preferred to extract multidimensional features from the preprocessed images through a convolutional neural network. Including the shape, color, texture, etc. of the cargo. The appearance features of the cargo are obtained by edge detection, contour extraction, and other methods. The color distribution and statistical information in the image are analyzed to extract features representing the color of the cargo. The surface texture features of the cargo are extracted using tools such as grayscale co-occurrence matrix and Gabor filter. The multidimensional features extracted from the acquired image are compared and matched with the preset cargo convolution feature set. The preset features that are most similar to the extracted features are found to determine the category of the cargo. The successfully matched multidimensional features are integrated to form a set of appearance features of the cargo.

[0022] S3: constructing a cargo sorting label library, and labeling the cargo appearance feature set based on the cargo sorting label library to obtain cargo sorting attribute label information.

[0023] Specifically, determine the classification of the label library, including sorting identification, sorting tools, and sorting warehouses, and configure a standardized data structure for each classification, including field name, data type, field length, etc. Ensure that the data structure can cover various sorting attributes and related information. You can build a database through a relational database such as MySQL, Oracle, or a NoSQL database such as MongoDB, configure the database table structure, and store various types of label information. Enter the initialized sorting label data into the database.

[0024] Use visual sensing equipment to obtain the appearance image of the goods. Apply image processing technology to extract the key appearance features of the goods, such as shape, color, size, etc. Match the extracted appearance features with the sorting logo in the label library. According to the matching results, assign corresponding sorting methods, sorting tools and other labels to the goods. Integrate the assigned label information to form complete cargo sorting attribute label information.

[0025] S4: Obtain target port distribution information, perform graph model abstraction on the target port distribution information, and generate a port path graph model.

[0026] Specifically, the geographical location information of the target port is collected, including longitude and latitude, coastline shape, berth location, etc., and the facility layout inside the port, such as warehouses, yards, loading and unloading areas, road networks, etc., is obtained. The collected geographical location information and facility layout data are integrated to ensure the accuracy and consistency of the data. The key factors that affect the path planning are extracted. The key locations in the port, such as warehouses, loading and unloading areas, and traffic nodes, are abstracted as nodes in the graph. Each node contains attributes such as location information, facility type, and capacity limit. According to the road network and transportation path of the port, the connection relationship between the nodes is abstracted as edges in the graph. Each edge contains attributes such as length, capacity, and transportation cost. The data structures in graph theory, such as adjacency matrix or adjacency list, are used to represent the relationship between nodes and edges. Ensure that the graph model can accurately reflect the actual layout and operating conditions of the port. The sorted port distribution information and graph model data are input into the path planning system. Create an instance of the graph model in the path planning system, including the creation and initialization of nodes and edges, and set parameters such as the initial state of nodes and the weight of edges. Verify the generated port path graph model to ensure its accuracy and completeness. Optimize the graph model according to actual conditions, such as adjusting node positions, modifying edge weights, etc., to improve the efficiency of path planning.

[0027] S5: performing sorting path planning for the cargo sorting attribute label information based on the port path graph model to obtain a target cargo sorting path, and automatically sorting the target loading and unloading cargo through the target cargo sorting path.

[0028] Specifically, according to the complexity and scale of the port path graph model, the sorting path is obtained according to the path planning algorithm, such as the Dijkstra algorithm, the A* search algorithm, or a more complex heuristic search algorithm. According to the cargo sorting attribute label information, such as cargo type, destination warehouse, priority, transportation tool, etc., the path planning goal is set, such as the shortest path, the fastest path, or the path with the lowest cost. Using the selected path planning algorithm, search for the optimal path from the current location of the cargo to the target location, such as a specific warehouse or loading and unloading area, on the graph model. After calculation by the path planning algorithm, determine an optimal sorting path, which should meet the set goals, such as the shortest, fastest, or lowest cost. Verify the calculated sorting path to ensure its feasibility in actual operation. Convert the planned sorting path into specific sorting instructions and send them to the automated sorting system. After receiving the instructions, automated sorting equipment, such as automatic guided vehicles AGV, robotic arms, etc., carry and sort the goods according to the planned path.

[0029] Furthermore, step S2 of the present application also includes:

[0030] Noise spectrum recognition is performed on the collected image information to obtain image noise feature information; filter feature parameter configuration is performed based on the image noise feature information to determine the image filter, and the collected image information is filtered and preprocessed by the image filter to obtain standard cargo image information; edge detection is performed on the standard cargo image information to determine cargo image edge information; foreground segmentation and marking is performed on the standard cargo image information based on the cargo image edge information to obtain cargo foreground image information; feature extraction and analysis is performed on the cargo foreground image information based on the preset cargo convolution feature set to obtain the cargo appearance feature set.

[0031] Specifically, the collected image information is subjected to spectrum analysis to identify the noise components in the image. The image is converted from the spatial domain to the frequency domain by Fourier transform and other methods, and then the spectrum characteristics of the noise are identified in the frequency domain. According to the results of the spectrum analysis, the characteristic information of the noise, such as the type, intensity, and distribution of the noise, is extracted. Based on the identified noise characteristic information, the parameters of the filter are configured. These parameters include the type of filter, such as low-pass, high-pass, band-pass, etc., cutoff frequency, order, etc. According to the configured filter parameters, a suitable filter, such as a Gaussian filter, a median filter, etc., is selected to filter the original image. The noise in the image is removed or weakened to improve the image quality. After filtering preprocessing, the standard cargo image information with most of the noise removed is obtained. The standard cargo image information is subjected to edge detection, and operators such as Sobel and Canny are used to identify the edges of objects in the image. After processing by the edge detection algorithm, the edge information of the cargo image is obtained. Based on the edge information of the cargo image, the foreground segmentation mark is performed on the standard cargo image. The foreground in the image, that is, the cargo, is distinguished from the background. After foreground segmentation marking, the image information containing only the cargo part, that is, the cargo foreground image information, is extracted. Using the convolutional neural network technology in deep learning, the foreground image information of the cargo is extracted based on the preset cargo convolution feature set. These features include key information such as the shape, texture, and color of the cargo. Through the feature extraction layer of the convolutional neural network, a comprehensive and compact cargo appearance feature set is finally obtained.

[0032] Further, if Figure 2 As shown, this application also includes:

[0033] According to the preset cargo convolution feature set, the collected image information is subjected to traversal convolution calculation to obtain an image convolution calculation result; the image convolution calculation result that meets the preset convolution threshold is used as image matching feature information; based on the image matching feature information, a multi-dimensional loss feature analysis is performed on the collected image information to determine the image multi-dimensional loss feature information; based on the image multi-dimensional loss feature information, the image matching feature information is subjected to different fusion to determine the cargo appearance feature set.

[0034] Specifically, the collected image information is traversed and convoluted using the preset cargo convolution feature set, and features matching the convolution kernel are extracted from the image. After the convolution calculation, a set of feature maps is obtained, which reflects the degree of matching between different positions in the image and the convolution kernel. The convolution threshold is set by a technician in this field to filter out features that respond strongly in the convolution calculation. Only when the response value of the feature map exceeds this threshold, the feature is considered to be a valid image matching feature. The image convolution calculation result that meets the preset convolution threshold is used as the image matching feature information. These feature information are the parts of the image that are highly matched with the preset convolution feature set. Multidimensional loss feature analysis is performed on the filtered image matching feature information. The loss of each feature in different dimensions is evaluated, for example, by calculating the similarity, difference or information entropy between the features. According to the results of the multidimensional loss feature analysis, those features with smaller losses in multiple dimensions are determined, that is, more robust and discriminative features. Based on the multidimensional loss feature information of the image, the image matching feature information is fused differently. Multiple features with different characteristics but complementary are fused together to form a more comprehensive and representative feature set. After the dissimilarity fusion, the feature set obtained is the cargo appearance feature set. This set contains the key features extracted from the collected images, which have been screened and fused, and can be used for tasks such as cargo identification, classification or sorting.

[0035] Furthermore, the present application also includes:

[0036] The collected image information is segmented according to the appearance application standard to obtain N cargo image regions; the collected image information is compared and analyzed based on the image matching feature information to obtain corresponding N regional feature difference information; key values ​​are assigned to the N cargo image regions to determine image region weight factor information; weighted calculation is performed on the N regional feature difference information according to the image region weight factor information to determine the image multi-dimensional loss feature information.

[0037] Specifically, according to the appearance application standard, including color, texture, shape and other features, the collected image information is segmented. After the segmentation is completed, N cargo image regions are obtained, each of which represents an independent part or cargo in the image. The image matching feature information is used to compare and analyze the N cargo image regions obtained by segmentation to find the difference between each region and the preset feature. After the comparison and analysis, the difference information between each cargo image region and the preset feature, that is, the regional feature difference information, is obtained. This information reflects the degree of mismatch between each region and the ideal feature. Each cargo image region is assigned a value according to its importance or criticality in the overall image. Based on the criticality assignment, the weight factor of each cargo image region is determined. Using the determined image region weight factor information, the N regional feature difference information is weightedly calculated, and the difference and importance of each region are comprehensively considered, so as to obtain a more accurate overall difference assessment. After the weighted calculation is completed, the result obtained is the multi-dimensional loss feature information of the image. This information comprehensively reflects the difference between the image and the preset features in various dimensions, and takes into account the importance of different regions.

[0038] Furthermore, step S3 of the present application also includes:

[0039] Sorting attributes are extracted from the cargo sorting label library to obtain a sorting attribute index set, and a cargo sorting label coordinate system is constructed according to the sorting attribute index set; the cargo sorting label coordinate system is divided into regions based on the cargo sorting label library to obtain a cargo feature sorting label region set; feature mapping is performed on the cargo appearance feature set based on the cargo sorting label coordinate system to generate a cargo appearance feature vector; regional label matching is performed according to the cargo appearance feature vector and the cargo feature sorting label region set to obtain the cargo sorting attribute label information.

[0040] Specifically, sorting-related attributes, such as sorting method, sorting mark, sorting tool, destination, etc., are extracted from the cargo sorting label library to form a sorting attribute index set. A multidimensional cargo sorting label coordinate system is constructed using the extracted sorting attribute index set. Based on the cargo sorting label library and the constructed sorting label coordinate system, the coordinate system is divided into regions. After the multidimensional coordinate system is divided by taking each attribute index as the coordinate axis of the cargo sorting label coordinate system, a cargo feature sorting label region set is obtained. Each region in this set is associated with a specific sorting label. The cargo appearance feature set is mapped to the cargo sorting label coordinate system. The cargo appearance features are associated with the sorting attributes. During the mapping process, each appearance feature is converted into a vector representation in the sorting label coordinate system, namely, the cargo appearance feature vector. This vector captures the relationship between the cargo appearance and the sorting attributes. The generated cargo appearance feature vector is used to match the cargo feature sorting label region set to find the sorting label region that best matches the cargo appearance feature. After the matching is completed, the sorting attribute label information corresponding to the appearance characteristics of the goods can be obtained. This information will be used to guide the sorting and classification of goods.

[0041] Furthermore, step S4 of the present application also includes:

[0042] Based on the target port distribution information, entity recognition is performed to obtain a port distribution entity set, and attribute classification and labeling are performed on the port distribution entity set to obtain a port entity attribute information set; graph model attributes are abstracted for the target port distribution information according to the port entity attribute information set to determine a graph model node set and a graph model edge set; based on the port entity attribute information set, model topology connection is performed on the graph model node set and the graph model edge set to generate the port path graph model.

[0043] Specifically, based on the target port distribution information, the port entities are identified through entity recognition technology, such as named entity recognition (NER) in natural language processing. These entities may include port name, location, type, and specific distribution facilities in the port. Specific distribution facilities include docks, roads, warehouses, etc. All identified port entities are organized into a set, namely, the port distribution entity set. Each entity in the port distribution entity set is classified and labeled. For example, each port entity can be labeled with its geographical location, scale, throughput, facility type, and other attributes. The classified and labeled port entities and their attribute information are organized into a set. According to the port entity attribute information set, the target port distribution information is abstracted by the attributes of the graph model. Each port entity is abstracted as a node in the graph model, and the connection relationship between nodes is defined according to the attribute information of the entity. Each port entity is abstracted as a graph model node, and all such nodes form a graph model node set. According to the relationship between port entities, specifically, docks, warehouses, etc. are abstracted as graph model nodes, and roads, railways, etc. are abstracted as connecting edges between nodes. These edges represent the connection relationship between ports and form a graph model edge set. The topological connection of the model is performed using the data in the port entity attribute information set and the previously defined graph model node set and edge set. That is, connections are established between nodes according to the defined edge relationships. After the topological connection of all nodes is completed, a complete port path graph model is formed. This model can intuitively display the connection relationship between ports.

[0044] Furthermore, step S5 of the present application also includes:

[0045] Iterative planning is performed on the port path graph model based on the cargo sorting attribute label information to obtain multiple cargo sorting path information; a sorting optimization target is defined, and a path fitness function is constructed according to the sorting optimization target; the multiple cargo sorting path information is evaluated and calculated using the path fitness function to determine multiple sorting path fitness information; the multiple cargo sorting path information is optimized based on the multiple sorting path fitness information to determine the target cargo sorting path.

[0046] Specifically, iterative planning is performed on the port path graph model based on the sorting attribute label information of the goods, such as sorting method, sorting warehouse and other labels. Through iterative planning, multiple possible cargo sorting path information is generated. Each path represents a possible route for the goods from the starting point to the destination. Before performing path optimization, the optimization goals, such as the shortest path, the lowest cost, the fastest time, etc., are clearly defined. These goals will serve as the basis for constructing the path fitness function. The path fitness function is a mathematical model used to evaluate the pros and cons of each path. It is defined based on sorting optimization goals, such as path length, transportation cost, time efficiency, etc., and a fitness value is calculated for each path. The path fitness function is used to evaluate and calculate multiple cargo sorting path information, and each cargo sorting path is evaluated. A specific fitness value is generated for each path, reflecting the degree of match between the path and the optimization goal. After the evaluation is completed, the fitness information of each path is obtained. This information will be used to compare and select the optimal path. According to the fitness information, multiple sorting paths are sorted and compared. The purpose of optimization is to find the path with the highest fitness, that is, the path that best meets the optimization goal. After optimization, the path with the highest fitness is selected as the target cargo sorting path. This path is considered to be the most suitable cargo sorting route under the conditions of satisfying the sorting attribute label information and the optimization goal.

[0047] Furthermore, the present application also includes:

[0048] The sorting optimization target is decomposed to obtain a set of sorting optimization sub-tasks; weights are assigned to the set of sorting optimization sub-tasks to determine sub-task weight assignment information; based on the sub-task weight assignment information, the set of sorting optimization sub-tasks is evaluated and fitted to construct the path fitness function.

[0049] Specifically, the sorting optimization goal is a comprehensive goal, such as improving sorting efficiency. This goal can be further decomposed into multiple specific subtasks, such as reducing transportation distance, reducing sorting time, and reducing sorting costs. All the decomposed subtasks are sorted into a set, namely the sorting optimization subtask set. Each subtask represents a specific aspect of the optimization goal. Different subtasks may have different importance in the overall optimization goal. Therefore, a weight value is assigned to each subtask according to the actual situation to reflect its relative importance in the overall optimization. After the weight assignment is completed, a subtask weight assignment information table or array is obtained, which contains each subtask and its corresponding weight value. Using the subtask weight assignment information, each subtask is quantitatively evaluated, an evaluation function or indicator is defined for each subtask, and weighted summation or other forms of fitting are performed according to the weight. Based on the evaluation and fitting results, a path fitness function is constructed. This function can comprehensively consider the influence of each subtask and calculate a comprehensive fitness value for each path according to the weight assignment information.

[0050] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

[0051] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application belong to the scope of the present application and its equivalent technology, the present application is also intended to include these modifications and variations.

Claims

1. A method for automated cargo sorting for port cargo loading and unloading, characterized in that: The method comprises: Obtain target cargo for loading and unloading through the port cargo management system, and obtain a preset cargo convolution feature set according to the appearance application standard of the target cargo for loading and unloading; Acquire the collected image information of the target loading and unloading cargo using a visual sensing device, perform multi-dimensional feature extraction on the collected image information based on the preset cargo convolution feature set, and obtain a cargo appearance feature set; Constructing a cargo sorting label library, and labeling the cargo appearance feature set based on the cargo sorting label library to obtain cargo sorting attribute label information; Obtaining target port distribution information, performing graph model abstraction on the target port distribution information, and generating a port path graph model; Based on the port path graph model, a sorting path is planned for the cargo sorting attribute label information to obtain a target cargo sorting path, and the target loading and unloading cargo is automatically sorted through the target cargo sorting path; The step of obtaining a set of appearance features of goods includes: Performing noise spectrum recognition on the collected image information to obtain image noise feature information; Based on the image noise characteristic information, filter characteristic parameters are configured to determine an image filter, and the collected image information is filtered and preprocessed by the image filter to obtain standard cargo image information; Performing edge detection on the standard cargo image information to determine cargo image edge information; Performing foreground segmentation and marking on the standard cargo image information based on the cargo image edge information to obtain cargo foreground image information; Based on the preset cargo convolution feature set, feature extraction and analysis are performed on the cargo foreground image information to obtain the cargo appearance feature set.

2. The method for automatic cargo sorting for port cargo loading and unloading according to claim 1, characterized in that: The step of obtaining the set of appearance features of the goods comprises: Performing traversal convolution calculation on the collected image information according to the preset cargo convolution feature set to obtain an image convolution calculation result; The image convolution calculation result that meets the preset convolution threshold is used as image matching feature information; Performing a multi-dimensional loss feature analysis on the collected image information based on the image matching feature information to determine the multi-dimensional loss feature information of the image; The image matching feature information is differentially fused based on the multi-dimensional loss feature information of the image to determine the appearance feature set of the goods.

3. The method for automated cargo sorting for port cargo loading and unloading according to claim 2, characterized in that: The determining of the multi-dimensional loss feature information of the image includes: Performing image segmentation on the collected image information according to the appearance application standard to obtain N cargo image regions; Compare and analyze the collected image information based on the image matching feature information to obtain corresponding N regional feature difference information; Assigning criticality to the N cargo image regions to determine image region weight factor information; The N region feature difference information is weightedly calculated according to the image region weight factor information to determine the image multi-dimensional loss feature information.

4. The method for automated cargo sorting for port cargo loading and unloading according to claim 1, characterized in that: The obtaining of cargo sorting attribute label information includes: Extracting sorting attributes from the cargo sorting label library, obtaining a sorting attribute index set, and constructing a cargo sorting label coordinate system according to the sorting attribute index set; Based on the cargo sorting label library, the cargo sorting label coordinate system is divided into regions to obtain a cargo feature sorting label region set; Performing feature mapping on the cargo appearance feature set based on the cargo sorting label coordinate system to generate a cargo appearance feature vector; The cargo sorting attribute label information is obtained by performing region label matching based on the cargo appearance feature vector and the cargo feature sorting labeling region set.

5. The method for automated cargo sorting for port cargo loading and unloading according to claim 1, characterized in that: The generating of the port route map model comprises: Perform entity recognition based on the target port distribution information to obtain a port distribution entity set, and classify and label the port distribution entity set to obtain a port entity attribute information set; Performing graph model attribute abstraction on the target port distribution information according to the port entity attribute information set, and determining a graph model node set and a graph model edge set; The graph model node set and the graph model edge set are topologically connected based on the port entity attribute information set to generate the port path graph model.

6. The method for automated cargo sorting for port cargo loading and unloading according to claim 1, characterized in that: The obtaining of the target cargo sorting path comprises: Iteratively planning the port path graph model based on the cargo sorting attribute label information to obtain multiple cargo sorting path information; Defining a sorting optimization goal, and constructing a path fitness function according to the sorting optimization goal; Using the path fitness function to evaluate and calculate the multiple cargo sorting path information, and determine the fitness information of the multiple sorting paths; The plurality of cargo sorting path information is optimized based on the plurality of sorting path fitness information to determine the target cargo sorting path.

7. The method for automated cargo sorting for port cargo loading and unloading according to claim 6, characterized in that: Constructing a path fitness function according to the sorting optimization objective includes: Decomposing the sorting optimization target to obtain a set of sorting optimization sub-tasks; Performing weight allocation on the sorting optimization sub-task set to determine sub-task weight allocation information; The sorting optimization sub-task set is evaluated and fitted based on the sub-task weight allocation information to construct the path fitness function.

Images