Automatic fish maw detecting and grading system

Abstract

The invention relates to the technical field of intelligent detection and quality grading, in particular to an automatic fish maw detection and grading system which integrates a spectral imaging module, a trace component analysis module, a physical property detection module and the like to realize comprehensive detection of external characteristics and internal components of fish maw. Through a multi-band light source and a high-resolution camera, an image from visible light to near-infrared light is obtained, and key indexes such as collagen, mineral substances and nutritional ingredients are accurately analyzed by utilizing a nondestructive testing technology; the feature extraction and fusion module adopts weighted mutual information and a self-adaptive penalty factor model to optimize features, the intelligent grading decision module provides accurate grading based on a feature intelligent discrimination method, and the self-adaptive control module dynamically adjusts equipment parameters to ensure the process to be stable. The system has the advantages of high efficiency, accuracy and intellectualization, and provides technical support for standardized grading and industrial production of the fish maw.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent detection and quality grading, and in particular to an automatic fish maw detection and grading system. Background Art

[0002] At present, the quality of fish maw is mainly determined through manual inspection and manual grading, which relies on the experience and subjective judgment of the inspectors. This traditional method is not only inefficient and costly, but also highly subjective and lacks a unified grading standard, which can easily lead to inconsistent grading results. In addition, manual testing is difficult to accurately and stably analyze the internal composition and physical properties of fish maw, and cannot meet the market's requirements for high quality and standardization.

[0003] In the existing technology, some studies have introduced technologies such as optical imaging, laser measurement and multispectral analysis to detect the appearance and structural characteristics of objects, but these technologies have limited application in fish maw detection. For example, optical imaging can capture the surface texture of fish maw, but it is difficult to provide in-depth composition information; while multispectral analysis technology can identify some material components, but the equipment is complex, the cost is high and the detection process is cumbersome. In addition, there has been some research on intelligent grading systems in food quality detection, such as fish and shellfish grading in aquatic products, but there is no automated grading system specifically for fish maw.

[0004] To sum up, an automated fish maw detection and grading system that can achieve rapid and accurate detection and standardized grading of fish maw, reduce manual intervention, improve detection efficiency and grading consistency, and meet the market demand for high-quality fish maw is a technical problem that needs to be solved urgently. Summary of the invention

[0005] In view of the above problems existing in the prior art, the present invention is proposed.

[0006] Therefore, the technical problem to be solved by the present invention is: through the combination of multi-module collaboration and intelligent algorithms, a new set of technical solutions is provided for fish maw quality detection and grading, which significantly improves the detection speed, grading accuracy and system reliability, and meets the needs of industrialization and large-scale production.

[0007] In order to solve the above technical problems, the present invention provides the following technical solutions: an automated fish maw detection and grading system, comprising: a spectral imaging module, a trace component analysis module, a physical property detection module, a feature extraction and fusion module, an intelligent grading decision module, an adaptive control module, and a data display and storage module; The spectral imaging module is responsible for obtaining multi-dimensional spectral image data of fish maw, capturing images of fish maw at different wavelengths through a multi-band light source and a high-resolution camera; achieving imaging from visible light to near-infrared range, obtaining subtle color differences, surface texture and structural features of fish maw, and eliminating the influence of ambient light; The spectral imaging module also includes a multi-band light source system, a high-resolution image sensor, an image preprocessing unit and a data synchronization unit; The multi-band light source system includes light sources with different wavelengths, supports adaptive adjustment, and selects the best band combination according to the surface reflection characteristics of the fish maw; The high-resolution image sensor includes a sensor for capturing the image of the fish maw after being irradiated with multi-band light and transmitting the image data to the processing system; The image preprocessing unit includes performing denoising, background correction and color standardization on the image data; The data synchronization unit includes synchronization with a multi-band light source system, a high-resolution image sensor and an image pre-processing unit.

[0008] As a preferred solution of the automated fish maw detection and grading system of the present invention, the trace component analysis module includes analyzing the content of trace components in the fish maw, including collagen, minerals and nutrients; the trace component analysis module combines chemical component detection and spectral analysis technology to detect the internal components of the fish maw in a non-destructive and rapid manner, specifically including near-infrared spectral analysis, Raman spectral detection, fluorescence spectral detection and trace sampling chemical analysis; The near infrared spectroscopy analysis includes irradiating the fish maw with near infrared spectroscopy technology, and identifying the content of collagen and water in the fish maw through the reflection and absorption characteristics of light at different wavelengths; The Raman spectrum detection includes providing molecular vibration information of fish maw to help identify its chemical composition; through laser irradiation, measuring the changes in Raman scattering spectrum, and detecting minerals and trace elements in fish maw; The fluorescence spectrum detection includes exciting specific molecules in the fish maw to generate fluorescence, and determining the content of the specific substance by analyzing the intensity and wavelength of the fluorescence; The micro-sampling chemical analysis is performed by combining chemical analysis techniques, including high performance liquid chromatography and gas chromatography-mass spectrometry, to detect trace components.

[0009] As a preferred solution of the automated fish maw detection and grading system of the present invention, the physical property detection module comprises using sensors and measurement technology to detect the physical characteristics of the fish maw, including weight detection, thickness measurement, density calculation, elasticity test and humidity measurement; The weight detection includes measuring the weight of the fish maw by a high-precision electronic balance and a sensor; The thickness measurement includes using a laser range finder sensor and an ultrasonic sensor to accurately measure the thickness of the fish maw, wherein the thickness is used to determine the content of nutritional components; The density calculation includes calculating the density of the fish maw by weight and volume, wherein the density is used to judge the quality of the fish maw, and if the density of the fish maw is high, the gelatin content is high; The elasticity test includes using a compression sensor and an elasticity detector to detect the elasticity of the fish maw, wherein the elasticity is used to judge the quality of the fish maw. If the elasticity of the fish maw is high, it is a high-quality fish maw; the detector determines the elasticity characteristics by applying pressure to the fish maw and measuring the ability of the fish maw to recover its deformation; The humidity measurement includes using a humidity sensor and a microwave humidity meter to detect the humidity content of fish maw, wherein humidity affects the storage and quality of fish maw.

[0010] As a preferred solution of the automated fish maw detection and grading system of the present invention, the feature extraction and fusion module includes extracting, analyzing and fusion of feature data of fish maw to provide feature information for grading decision; The feature extraction includes image feature extraction, physical feature extraction and component feature extraction, and performs feature selection, dimensionality reduction and feature fusion; The image feature extraction includes shape features, texture features and color features; The feature selection and dimensionality reduction include screening fish maw features using statistical analysis and feature selection algorithms; The feature fusion includes integrating the fish maw features after screening and dimension reduction into a unified feature vector, wherein the fusion includes feature splicing, weighted fusion and multi-layer perception fusion.

[0011] As a preferred solution of the automated fish maw detection and grading system described in the present invention, the intelligent grading decision module includes judging the quality grade of the obtained characteristic data, making a grading decision, and analyzing the fish maw characteristic data using a characteristic intelligent discrimination method; The quality grade judgment includes evaluating the quality level of fish maws by fish maw characteristics, and classifying fish maws into high-quality, medium and low-quality according to preset standards; The grading decision includes further determining the processing method of the fish maw based on the quality grade judgment result, making a decision according to the fish maw quality grade and grading standard, and determining the processing flow and placement location of the fish maw sample.

[0012] As a preferred solution of the automated fish maw detection and grading system described in the present invention, the adaptive control module automatically adjusts the control parameters of the equipment according to the quality grade of the fish maw and the grading decision result, so as to realize the automated grading placement and processing of the fish maw; dynamically adjusts the operation process and parameters of the system by receiving the output information of the intelligent grading decision module in real time; The adaptive control module also includes an automatic triggering of alarm and pause mechanisms when equipment failures and abnormal conditions are detected.

[0013] As a preferred solution of the automated fish maw detection and grading system described in the present invention, the data display and storage module includes a visual display of the fish maw detection results and grading decision results, and performs data management and storage to facilitate data backup and recovery; The data management and storage includes automatically storing the fish maw detection results and grading decision results in a database; The data backup and recovery is an automatic backup function equipped in the data display and storage module, which saves data to local backup and cloud servers.

[0014] Another object of the present invention is to provide an automated fish maw detection and grading method, which realizes effective screening, optimization and intelligent decision-making of multi-dimensional feature data of fish maw through feature selection algorithm and characteristic intelligent discrimination method. The method combines the weighted mutual information feature selection model and the adaptive penalty factor fusion model, dynamically adjusts the feature weight based on the set threshold, accurately extracts the features that contribute most to the classification, and eliminates the influence of redundant features and outliers; through feature impact scoring and weight optimization technology, calculates and selects the maximum value of the feature's influence on the target classification label, optimizes the efficiency and accuracy of the classification model, and provides scientific and intelligent decision-making support for fish maw quality grading.

[0015] In order to solve the above technical problems, the present invention provides the following technical solutions: an automated fish maw detection and grading system, comprising: a feature selection algorithm and a characteristic intelligent discrimination method; The feature selection algorithm includes a weighted mutual information feature selection model and an adaptive penalty factor feature fusion model; wherein, the fish maw feature data set is: ; Defining Characteristics Weighted mutual trust score for: in, Representation characteristics and target classification label The mutual information between Representation characteristics With features The mutual information of and represents the weight factor, Represents the total number of features in the feature dataset, Represents the feature index used for the mutual redundancy part in the weighted mutual information calculation; Introducing adaptive penalty factor Dynamically adjust feature weights to define the final score of the feature for: in, Characteristic norm; represents the standard deviation of the feature; represents the mean of the feature; represents the adjustment coefficient; By selecting the preset value, adjust and The value of can be adjusted to meet the classification requirements by adjusting the weight; if the value is increased If the value is large, we should pay attention to the effectiveness of the feature. value, the redundancy of features is reduced; The weighted mutual trust score and final score of the features are calculated in real time. The system automatically screens and compares and marks the optional features. The threshold is set to ,like , then the feature is marked as a valid feature, if , then the feature is marked as redundant; let the threshold be ,like , then the feature is retained as an optional feature. If , then exclude the feature; when it is a dual threshold mode, the system gives priority to satisfying , and further filter When the single threshold mode is used, the system uses a single threshold for comparison and screening; using the adaptive penalty factor introduced Adjust the system penalty factor and set it to ,when When , it is marked as a high penalty feature, indicating low contribution to the model and redundancy, so the system penalty factor is increased to further screen features; when When , it is marked as a low penalty feature, indicating that it has analytical value and importance to the model, so the system penalty factor is reduced and the feature is introduced; The characteristic intelligent identification method includes characteristic impact scoring and classification decision, wherein the characteristic impact scoring is defined , directly related features For the final classification Impact: in, Represents a given category Lower Features The conditional probability of Representation characteristics The marginal probability of Function is used to amplify the probability ratio; feature impact score directly calculates the predicted category : in, Representation characteristics The weight of represents the final classification decision, Represents the category index, The feature index representing the score calculation; Set the classification weight threshold to ,when When , the feature is judged as an important feature and participates in the final classification decision; The final classification decision is updated as: .

[0016] A computer device comprises a memory and a processor, wherein the memory stores a computer program, and wherein the processor implements the steps of an automated fish maw detection and grading system as described above when executing the computer program.

[0017] A computer-readable storage medium having a computer program stored thereon, characterized in that when the computer program is executed by a processor, the steps of an automated fish maw detection and grading system as described above are implemented.

[0018] Beneficial effects of the present invention: The present invention provides an automated fish maw detection and grading system, which has the significant advantages of high efficiency, precision and intelligence. By integrating modules such as spectral imaging, trace component analysis, and physical property detection, the detection requirements of the external characteristics and internal components of fish maw are fully covered. By using multi-band light sources and high-resolution cameras, image acquisition from visible light to near-infrared is achieved, and key indicators such as collagen, minerals and nutrients are accurately analyzed through non-destructive testing technology. The feature extraction and fusion module combines weighted mutual information feature selection and adaptive penalty factor model to screen and optimize features to generate efficient and unified feature vectors. The intelligent grading decision module provides accurate quality grading based on feature scoring and weight optimization technology through characteristic intelligent discrimination methods. The adaptive control module dynamically adjusts the equipment operating parameters to ensure the stability and adaptability of the grading process, and combines the data display and storage module to realize the visualization and safe storage of the detection results. The present invention not only improves the efficiency and accuracy of fish maw detection and grading through the combination of intelligent algorithms and automation technologies, but also meets the needs of industrialization and large-scale production, and provides important technical support for standardized grading. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0020] Figure 1 A module diagram of an automated fish maw detection and grading system provided for one embodiment of the present invention.

[0021] Figure 2 A visualization diagram of feature importance of an automated fish maw detection and grading system provided by one embodiment of the present invention.

[0022] Figure 3 A comparison chart of classification accuracy of an automated fish maw detection and grading system provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0023] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.

[0024] Example 1, reference Figure 1, is an embodiment of the present invention, which provides an automated fish maw detection and grading system, comprising: Spectral imaging module, trace component analysis module, physical property detection module, feature extraction and fusion module, intelligent grading decision module, adaptive control module and data display and storage module.

[0025] It should be noted that the spectral imaging module is responsible for obtaining multi-dimensional spectral image data of fish maw, capturing images of fish maw at different wavelengths through multi-band light sources and high-resolution cameras; achieving imaging from visible light to near-infrared range, obtaining subtle color differences, surface texture and structural characteristics of fish maw, and eliminating the influence of ambient light; the trace component analysis module includes analyzing the content of trace components in fish maw, including collagen, minerals and nutrients; the trace component analysis module combines chemical component detection and spectral analysis technology to detect the internal components of fish maw in a non-destructive and rapid manner, specifically including near-infrared spectral analysis, Raman spectral detection, fluorescence spectral detection and trace sampling chemical analysis; the physical property detection module includes using sensors and measurement technology to detect the physical characteristics of fish maw, including weight inspection The module includes the following: measurement, thickness measurement, density calculation, elasticity test and humidity measurement; the feature extraction and fusion module includes feature extraction, analysis and fusion of the feature data of fish maw, and provides feature information for grading decision; the intelligent grading decision module includes quality grade judgment of the obtained feature data, making grading decision, and analyzing the fish maw feature data by using the characteristic intelligent discrimination method; the adaptive control module includes automatically adjusting the control parameters of the equipment according to the quality grade of the fish maw and the grading decision results, so as to realize the automatic grading placement and processing of the fish maw; the output information of the intelligent grading decision module is received in real time, and the operation process and parameters of the system are dynamically adjusted; the data display and storage module includes visual display of the fish maw detection results and grading decision results, and data management and storage, so as to facilitate data backup and recovery; This system aims to achieve comprehensive detection and grading of fish maw from external features to internal components through the collaborative work of multiple modules; the spectral imaging module captures multi-dimensional spectral images of fish maw in the visible to near-infrared wavelength range through a multi-band light source system and a high-resolution camera, and combines with the image preprocessing unit to perform denoising, background correction and color standardization to eliminate ambient light interference and ensure the stability and accuracy of image data; the multi-band light source system of this module supports adaptive selection of the best band combination according to the surface reflection characteristics of the fish maw to optimize the imaging effect, thereby obtaining high-quality images of the surface texture, color differences and structural characteristics of the fish maw.

[0026] Furthermore, the trace component analysis module combines near-infrared spectroscopy, Raman spectroscopy, fluorescence spectroscopy and chemical analysis technology to quickly analyze key indicators such as collagen, minerals and nutrients in fish maw through non-destructive testing; near-infrared spectroscopy technology uses the reflection and absorption characteristics of light to identify the collagen and moisture content inside fish maw, Raman spectroscopy provides high-sensitivity support for mineral and trace element detection through molecular vibration information, and fluorescence spectroscopy analyzes protein and amino acid content by exciting molecular fluorescence; in addition, this module combines trace sampling chemical analysis technology (such as high-performance liquid chromatography and gas chromatography-mass spectrometry) to further improve the accuracy and breadth of detection.

[0027] Furthermore, the intelligent grading decision module uses the characteristic intelligent identification method to achieve accurate evaluation and scientific grading of fish maw quality, providing an efficient and feasible solution for industrial application; the core of the module is based on the feature impact scoring And weight optimization model, through accurate feature screening and dynamic weight adjustment, to achieve diverse grading needs; In the actual fish maw production and grading scenarios, the quality of fish maw will be affected by the combined effects of physical properties (such as thickness and density) and chemical composition (such as collagen and mineral content); therefore, the selection and weight distribution of features are directly related to the accuracy of grading; for example, collagen content and elasticity may be the core concerns, while for ordinary processing requirements, high-end fish maw grading with humidity and heavy taste standards, the uniformity of the quantity may be more important; the module influences the scoring formula through features: Calculate the relative contribution of features in the classification task, where represents the conditional probability of a feature under a specific classification, Represents its marginal probability. This formula can intuitively reflect the distinguishability of the feature to the classification results; Specifically, weight threshold interactive setting: the user enters the weight threshold through the interface The system uses this as a benchmark to screen important features. If a feature weight , the system will mark the feature as an important feature and participate in the final classification calculation; for example, in the analysis of collagen content, if its weight is higher than the threshold, the system will give priority to its impact on the classification and use it as the main decision basis; on the contrary, if the humidity weight is , the system reduces its influence in the classification to avoid interfering with key decisions.

[0028] Dynamic weight optimization: In batch fish maw testing, samples from different batches may have different distribution characteristics; by adjusting feature weights in real time , the system can dynamically respond to changes between features; for example, when the thickness feature of a batch of samples has significant differences in distribution (such as low mean and large standard deviation), the system will automatically increase the thickness weight to ensure the fairness and scientificity of the grading; Visual feedback interaction: The system provides visualization functions for feature importance ranking and impact weights, and displays the classification contribution of features through bar charts or heat maps. Users can manually fine-tune the weight priority of features according to the actual needs of the grading task. For example, in the inspection of export products, users can prioritize the weight of colloid content and elasticity data to meet the high standards of the market.

[0029] The intelligent grading module also supports adaptive learning of weight adjustment. In continuous batch production, the system dynamically adjusts the initial weight parameters by analyzing historical data. For example, when humidity frequently appears in an abnormal range (such as the standard deviation exceeds a specific value), the system will gradually reduce the impact of its weight on grading, and vice versa. This dynamic learning mechanism improves the system's adaptability in complex production environments. For example, in the detection of high-end fish maw for export, if the gelatin content of a batch of samples is widely distributed (such as a mixture of high-end and low-end products), the system first collects the multi-dimensional feature data of the samples in a comprehensive manner, including core indicators such as thickness, density, collagen content and elasticity; then, through the feature scoring function: The system calculates the contribution of each feature to the classification and , dynamically optimize the importance of features; this process can prioritize key features (such as colloid content) as important features when samples are unevenly distributed, and enhance their influence in the classification model; In order to deal with the possible interference of minor features in the sample (such as humidity and color difference) on the grading results, the system will introduce feature weight thresholds , filter out redundant features that contribute less to classification and reduce their weights; for example, when the humidity distribution deviates from the standard range and affects the classification accuracy of the colloid content, the system reduces the humidity weight to weaken its influence in classification, ensuring that the classification results focus on the core features; In addition, the system also has a real-time optimization and feedback mechanism. During the detection process, the system can dynamically adjust the weight parameters according to the feature distribution of each batch of samples. For example, when the mean of the density feature of a batch of samples is low and the variance is large, the system automatically increases the density weight to strengthen the influence of the feature and updates the classification model in real time. In addition, through the visual interface, users can intuitively view the weight distribution of each feature and fine-tune the weight parameters through sliders or numerical input. This interactive design not only improves the flexibility of the system, but also enhances the user's controllability of the classification process. In addition, in large-scale production scenarios, the intelligent grading module can also be linked with the adaptive control module to realize the automation of the grading process; for each batch of fish maw samples detected, the system dynamically adjusts the equipment parameters according to the real-time grading results; for example, when a batch of high-quality fish maw samples is detected, the system adjusts the sorting speed and placement of the grading equipment to ensure that the high-quality samples enter the high-end product sequence; for low-end samples, the system optimizes the sorting strategy and allocates them to the sequence for processing purposes; Through the above-mentioned dynamic optimization and interactive design, the system can effectively improve the classification accuracy; for example, in the high-standard export inspection, the gelatin content and elasticity are marked as core features by the system, and their weights are significantly higher than those of humidity and color features; this optimization strategy can not only ensure that the sample grading results meet the quality requirements of the international market, but also reduce the complexity and error rate of detection, and provide reliable grading support for production companies; in summary, the intelligent grading module provides a realistic and feasible solution for the efficient and accurate grading of fish maw through dynamic feature weight optimization, flexible interactive design and automated control capabilities.

[0030] The above is a schematic scheme of an automated fish maw detection and grading system of this embodiment. It should be noted that the technical scheme of this automated fish maw detection and grading system and the technical scheme of the above-mentioned automated fish maw detection and grading system belong to the same concept. The details of the technical scheme of an automated fish maw detection and grading method in this embodiment that are not described in detail can all be referred to the description of the technical scheme of the above-mentioned automated fish maw detection and grading system.

[0031] An automated fish maw detection and grading system in this embodiment is characterized by comprising: a feature selection algorithm and a characteristic intelligent discrimination method; The feature selection algorithm includes a weighted mutual information feature selection model and an adaptive penalty factor feature fusion model; wherein, the fish maw feature data set is: ; Defining Characteristics Weighted mutual trust score for: in, Representation characteristics and target classification label The mutual information between Representation characteristics With features The mutual information of represents the redundancy between features; and Represents the weight factor, which is used to balance the feature contribution and redundancy. By optimizing the weight coefficient, the bias of feature selection is dynamically adjusted. Represents the total number of features in the feature dataset, i.e. the fish maw feature dataset The number of features contained in represents the feature index used for the mutual redundancy part in the weighted mutual information calculation, The value range is 1 to , and satisfies , indicating the current calculation characteristics With characteristics other than itself The redundancy between Introducing adaptive penalty factor Dynamically adjust feature weights to define the final score of the feature for: in, Characteristic Norm, reflects the size of eigenvalue; represents the standard deviation of the feature; represents the mean of the feature; Represents the adjustment coefficient, set to ; By selecting the preset value, adjust and The value of can be adjusted to meet the classification requirements by adjusting the weight; if the value is increased If the value is large, we should pay attention to the effectiveness of the feature. value, the redundancy of features is reduced; The weighted mutual trust score and final score of the features are calculated in real time. The system automatically screens and compares and marks the optional features. The threshold is set to ,like , then the feature is marked as a valid feature, if , then the feature is marked as redundant; let the threshold be ,like , then the feature is retained as an optional feature. If , then exclude the feature; when it is a dual threshold mode, the system gives priority to satisfying , and further filter When the single threshold mode is used, the system uses a single threshold for comparison and screening; using the adaptive penalty factor introduced Adjust the system penalty factor and set it to ,when When , it is marked as a high penalty feature, indicating low contribution to the model and redundancy, so the system penalty factor is increased to further screen features; when When , it is marked as a low penalty feature, indicating that it has analytical value and importance to the model, so the system penalty factor is reduced and the feature is introduced; The characteristic intelligent identification method includes characteristic impact scoring and classification decision, wherein the characteristic impact scoring is defined , directly related features For the final classification Impact: in, Represents a given category Lower Features The conditional probability of Representation characteristics The marginal probability of Function is used to amplify the probability ratio; feature impact score directly calculates the predicted category : in, Representation characteristics The weight of represents the final classification decision, Represents the category index, The feature index value range for score calculation is , used to identify the feature currently being examined ; Set the classification weight threshold to ,when When , the feature is judged as an important feature and participates in the final classification decision; The final classification decision is updated as: .

[0032] This embodiment also provides a computing device, applicable to an automated fish maw detection and grading system, including: Memory and processor; the memory is used to store computer executable instructions, and the processor is used to execute computer executable instructions to implement an automated fish maw detection and grading system as proposed in the above embodiment. This embodiment also provides a storage medium on which a computer program is stored. When the program is executed by a processor, an automated fish maw detection and grading system as proposed in the above embodiment is implemented. The storage medium proposed in this embodiment and the automated fish maw detection and grading system proposed in the above embodiment belong to the same inventive concept. The technical details not fully described in this embodiment can be found in the above embodiment, and this embodiment has the same beneficial effects as the above embodiment. If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc., which can store program codes.

[0033] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in conjunction with such instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in conjunction with such instruction execution systems, devices or apparatuses.

[0034] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0035] Example 2, reference Figure 2-Figure 3 , which is an embodiment of the present invention, provides an automated fish maw detection and grading system. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through economic benefit calculation and simulation experiments.

[0036] Subjects: The 1,000 fish maw samples were divided into three categories: high-end, mid-range and low-end, and the samples were evenly distributed.

[0037] Five core characteristics were extracted for each sample: thickness, density, collagen content, elasticity, and moisture.

[0038] Experimental Group: The experimental group adopted the automated fish maw detection and grading system of the present invention, and the feature selection algorithm was based on a weighted mutual information and adaptive penalty factor fusion model.

[0039] Dynamically adjust feature weights and set feature impact score thresholds , important features participate in the final classification decision.

[0040] Control group: The control group used traditional manual grading without feature selection and relied on manual experience judgment.

[0041] Only a few features (thickness and collagen content) are used as the basis for grading.

[0042] Experimental steps: Feature extraction: The spectral imaging module is used to obtain multidimensional image data of the sample, and the image preprocessing unit is used to perform denoising, background correction and standardization.

[0043] Extract core features and record the feature values ​​for each sample.

[0044] Grouping: 1000 samples were randomly divided into an experimental group and a control group, with 500 samples in each group.

[0045] Feature screening and optimization (experimental group): Calculate the weighted mutual trust score for each feature and final score .

[0046] Based on the scoring threshold and Filter valid features.

[0047] Dynamically adjust the penalty factor to optimize important features based on adaptive weights.

[0048] Grading and discrimination: Experimental group: Calculate feature impact scores based on feature intelligent identification method , optimize classification weights , perform hierarchical decision making.

[0049] Control group: Manual grading based on experience and traditional rules.

[0050] Results recorded: Record the grading results, accuracy and processing time of the experimental group and the control group. The experimental data are shown in Table 1 below: Table 1 Data record table Experimental results analysis: The accuracy of the experimental group: By optimizing feature selection, the experimental group achieved a classification accuracy of 95.2%.

[0051] After dynamically adjusting the weights, the contribution of important features (such as collagen content) to the results is significantly improved.

[0052] The accuracy of the control group: The control group relied on experience and a small number of features, with a classification accuracy of only 82.4%, and its judgment of the internal composition of the samples was inaccurate.

[0053] Processing time: The experimental group took 30 minutes to process 500 samples.

[0054] The control group required manual participation and the processing time was 120 minutes.

[0055] The experimental results show that the automated fish maw detection and grading system of the present invention significantly improves the accuracy and efficiency of grading through feature screening and dynamic optimization, and solves the problems of low accuracy and poor efficiency in traditional grading methods. This shows that the system is suitable for large-scale production scenarios and can meet the needs of the high-end fish maw market.

[0056] As shown in the figure: like Figure 2 As shown: The data comes from the weighted mutual information feature selection model and the adaptive penalty factor feature fusion model described in the document.

[0057] The weight of each feature (e.g., thickness, density, collagen characteristics, elasticity, and moisture) is calculated by weighted mutual information scoring and the final score after adaptive adjustment Calculated.

[0058] By setting the threshold and Screen the features, retain efficient features, and exclude redundant features. Figure 2 It can be seen that collagen features have the highest importance (weight of 0.35), followed by elasticity and thickness, while humidity and density have relatively low contributions to classification.

[0059] Figure 2 How to obtain data: The characteristic data is collected in real time through the spectral imaging module, physical property detection module and trace component analysis module.

[0060] After the model is used to score the data, the features with higher scores are retained, and the features with lower scores are weighted down or removed.

[0061] The data shown in the figure comes directly from the feature weight values ​​of the model, and a bar chart is drawn after normalization.

[0062] like Figure 3 As shown: The figure shows the classification accuracy of the experimental group (using the feature intelligent discrimination method) and the control group (traditional classification method) at different classification levels (high level, medium level, and low level).

[0063] The calculation formula for classification accuracy is: Accuracy = number of correctly classified samples / total number of samples The formula used in the experimental group is that the above-mentioned features affect the scoring and classification decisions.

[0064] Figure 3 How to obtain data: Experimental Group: Using the above intelligent discrimination algorithm, the features are weighted on multiple test samples, and the classification efficiency is improved by dynamically adjusting the weight coefficients.

[0065] The data comes from the actual operation results of the test system on samples of different categories.

[0066] Control group: Using traditional fixed rules or classification methods based on simple linear models, the data is relatively dependent on the individual indicators of the features.

[0067] The classification results of the control group are somewhat subjective and may overlook some important features.

[0068] Figure 3 Data analysis: High level categories: The accuracy of the experimental group was higher than 0.9, while that of the control group was only around 0.8, indicating that the intelligent discrimination method has a more accurate grasp of the characteristics of advanced samples.

[0069] Medium-level category: The accuracy of the experimental group remained above 0.85, while the control group was slightly lower, about 0.7-0.75 Low-level category: The experimental group still performs better than the control group in low-level samples (about 0.9 vs. 0.8).

[0070] Figure 2 Conclusions from the data: The experimental group significantly improved the classification accuracy through the intelligent grading algorithm, especially in the high-level and low-level categories.

[0071] Since the control group failed to dynamically adjust the feature weights, the classification results were slightly insufficient in accuracy.

[0072] This experiment shows that the intelligent grading algorithm has obvious advantages in practical application scenarios, can adapt to complex sample distribution, and improve the overall performance of the grading system.

[0073] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. An automated fish maw detection and grading system, characterized by: include: Spectral imaging module, trace component analysis module, physical property detection module, feature extraction and fusion module, intelligent grading decision module, adaptive control module and data display and storage module; The spectral imaging module is responsible for obtaining multi-dimensional spectral image data of fish maw, capturing images of fish maw at different wavelengths through a multi-band light source and a high-resolution camera; achieving imaging from visible light to near-infrared range, obtaining subtle color differences, surface texture and structural features of fish maw, and eliminating the influence of ambient light; The spectral imaging module also includes a multi-band light source system, a high-resolution image sensor, an image preprocessing unit and a data synchronization unit; The multi-band light source system includes light sources with different wavelengths, supports adaptive adjustment, and selects the best band combination according to the surface reflection characteristics of the fish maw; The high-resolution image sensor includes a sensor for capturing the image of the fish maw after being irradiated with multi-band light and transmitting the image data to the processing system; The image preprocessing unit includes performing denoising, background correction and color standardization on the image data; The data synchronization unit includes synchronization with a multi-band light source system, a high-resolution image sensor and an image pre-processing unit.

2. The automated fish maw detection and grading system according to claim 1, characterized in that: The trace component analysis module includes analyzing the content of trace components in fish maw, including collagen, minerals and nutrients; the trace component analysis module combines chemical component detection and spectral analysis technology to detect the internal components of fish maw in a non-destructive and rapid manner, specifically including near-infrared spectral analysis, Raman spectral detection, fluorescence spectral detection and trace sampling chemical analysis; The near infrared spectroscopy analysis includes irradiating the fish maw with near infrared spectroscopy technology, and identifying the content of collagen and water in the fish maw through the reflection and absorption characteristics of light at different wavelengths; The Raman spectrum detection includes providing molecular vibration information of fish maw to help identify its chemical composition; through laser irradiation, measuring the changes in Raman scattering spectrum, and detecting minerals and trace elements in fish maw; The fluorescence spectrum detection includes exciting specific molecules in the fish maw to generate fluorescence, and determining the content of the specific substance by analyzing the intensity and wavelength of the fluorescence; The micro-sampling chemical analysis is performed by combining chemical analysis techniques, including high performance liquid chromatography and gas chromatography-mass spectrometry, to detect trace components.

3. The automated fish maw detection and grading system according to claim 2, characterized in that: The physical property detection module includes using sensors and measurement technology to detect the physical characteristics of the fish maw, including weight detection, thickness measurement, density calculation, elasticity test and humidity measurement; The weight detection includes measuring the weight of the fish maw by a high-precision electronic balance and a sensor; The thickness measurement includes using a laser range finder sensor and an ultrasonic sensor to accurately measure the thickness of the fish maw, wherein the thickness is used to determine the content of nutritional components; The density calculation includes calculating the density of the fish maw by weight and volume, wherein the density is used to judge the quality of the fish maw, and if the density of the fish maw is high, the gelatin content is high; The elasticity test includes using a compression sensor and an elasticity detector to detect the elasticity of the fish maw, wherein the elasticity is used to judge the quality of the fish maw. If the elasticity of the fish maw is high, it is a high-quality fish maw; the detector determines the elasticity characteristics by applying pressure to the fish maw and measuring the ability of the fish maw to recover its deformation; The humidity measurement includes using a humidity sensor and a microwave humidity meter to detect the humidity content of fish maw, wherein humidity affects the storage and quality of fish maw.

4. The automated fish maw detection and grading system according to claim 3, characterized in that: The feature extraction and fusion module includes feature extraction, analysis and fusion of the feature data of fish maw to provide feature information for grading decision; The feature extraction includes image feature extraction, physical feature extraction and component feature extraction, and performs feature selection, dimensionality reduction and feature fusion; The image feature extraction includes shape features, texture features and color features; The feature selection and dimensionality reduction include screening fish maw features using statistical analysis and feature selection algorithms; The feature fusion includes integrating the fish maw features after screening and dimension reduction into a unified feature vector, wherein the fusion includes feature splicing, weighted fusion and multi-layer perception fusion.

5. The automated fish maw detection and grading system according to claim 4, characterized in that: The intelligent grading decision module includes judging the quality grade of the obtained characteristic data, making a grading decision, and analyzing the fish maw characteristic data using a characteristic intelligent discrimination method; The quality grade judgment includes evaluating the quality level of fish maws by fish maw characteristics, and classifying fish maws into high-quality, medium-quality and low-quality according to preset standards; The grading decision includes further determining the processing method of the fish maw based on the quality grade judgment result, making a decision according to the fish maw quality grade and grading standard, and determining the processing flow and placement location of the fish maw sample.

6. The automated fish maw detection and grading system according to claim 5, characterized in that: The adaptive control module automatically adjusts the control parameters of the equipment according to the quality grade of the fish maw and the grading decision results, so as to realize the automatic grading placement and processing of the fish maw; dynamically adjusts the operation process and parameters of the system by receiving the output information of the intelligent grading decision module in real time; The adaptive control module also includes an automatic triggering of alarm and pause mechanisms when equipment failures and abnormal conditions are detected.

7. The automated fish maw detection and grading system according to claim 6, characterized in that: The data display and storage module includes visual display of fish maw detection results and grading decision results, and performs data management and storage to facilitate data backup and recovery; The data management and storage includes automatically storing the fish maw detection results and grading decision results in a database; The data backup and recovery is an automatic backup function equipped in the data display and storage module, which saves data to local backup and cloud servers.

8. An automated fish maw detection and grading system according to any one of claims 1 to 7, characterized in that: include: Feature selection algorithms and intelligent feature identification methods; The feature selection algorithm includes a weighted mutual information feature selection model and an adaptive penalty factor feature fusion model; wherein, the fish maw feature data set is: ; Defining Characteristics Weighted mutual trust score for: in, Representation characteristics and target classification label The mutual information between Representation characteristics With features The mutual information of and represents the weight factor, Represents the total number of features in the feature dataset, Represents the feature index used for the mutual redundancy part in the weighted mutual information calculation; Introducing adaptive penalty factor Dynamically adjust feature weights to define the final score of the feature for: in, Characteristic norm; represents the standard deviation of the feature; represents the mean of the feature; represents the adjustment coefficient; By selecting the preset value, adjust and The value of can be adjusted to meet the classification requirements by adjusting the weight; if the value is increased If the value is large, we should pay attention to the effectiveness of the feature. value, the redundancy of features is reduced; The weighted mutual trust score and final score of the features are calculated in real time. The system automatically screens and compares and marks the optional features. The threshold is set to ,like , then the feature is marked as a valid feature, if , then the feature is marked as redundant; let the threshold be ,like , then the feature is retained as an optional feature. If , then exclude the feature; when it is a dual threshold mode, the system gives priority to satisfying , and further filter When the single threshold mode is used, the system uses a single threshold for comparison and screening; using the adaptive penalty factor introduced Adjust the system penalty factor and set it to ,when When , it is marked as a high penalty feature, indicating low contribution to the model and redundancy, so the system penalty factor is increased to further screen features; when When , it is marked as a low penalty feature, indicating that it has analytical value and importance to the model, so the system penalty factor is reduced and the feature is introduced; The characteristic intelligent identification method includes characteristic impact scoring and classification decision, wherein the characteristic impact scoring is defined , directly related features For the final classification Impact: in, Represents a given category Lower Features The conditional probability of Representation characteristics The marginal probability of Function is used to amplify the probability ratio; feature impact score directly calculates the predicted category : in, Representation characteristics The weight of represents the final classification decision, Represents the category index, The feature index representing the score calculation; Set the classification weight threshold to ,when When , the feature is judged to be an important feature and participates in the final classification decision; The final classification decision is updated as: 。 9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of an automated fish maw detection and grading system as described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of an automated fish maw detection and grading system as described in any one of claims 1 to 7 are implemented.

Images