A quail egg intelligent detection method and system based on multi-source data fusion

Abstract

The application discloses a quail egg intelligent detection method and system based on multi-source data fusion, and relates to the technical field of quail egg quality detection. The quail egg intelligent detection method based on multi-source data fusion extracts image texture features and spectral response features by acquiring visible light images, near-infrared images and geometric parameter data of quail eggs, analyzes image equalization indexes and structure transmission indexes respectively, analyzes geometric morphological indexes in combination with morphological parameters, and constructs comprehensive state indexes, thereby determining the quality grade of the quail eggs to be detected. The application extracts image equalization indexes, structure transmission indexes and geometric morphological indexes by fusing visible light images, near-infrared images and geometric parameters, constructs comprehensive state indexes, can comprehensively evaluate the quality of quail eggs from multiple dimensions such as appearance, light transmission and morphology, effectively identifies internal abnormal conditions that are prone to be missed in traditional image detection, and improves the accuracy of quality grade judgment and the stability of overall detection results.

Description

A Smart Detection Method and System for Quail Eggs Based on Multi-Source Data Fusion Technical Field

[0001] This invention relates to the field of quail egg quality detection technology, specifically to an intelligent quail egg detection method and system based on multi-source data fusion. Background Technology

[0002] Quail eggs, as a common type of poultry egg, have a quality that directly affects consumer health and market circulation efficiency. During egg grading, selection, and factory inspection, it is often necessary to assess key factors such as appearance integrity, internal structure, and morphological parameters. Traditional testing methods mostly rely on visual inspection or single image recognition, typically judging appearance features such as cracks and stains based on visible light images. Other methods incorporate near-infrared transmission imaging or dimensional measurement equipment to analyze the egg's translucency and geometric parameters.

[0003] The limitations of existing technologies include at least the following problems: Current quail egg detection methods typically rely on images or physical parameters from a single source for quality assessment, making it difficult to conduct a unified analysis of multiple key dimensions such as egg surface texture, shell light transmittance, and geometric morphology. This leads to biased and distorted quality assessment results. For example, while visual identification based solely on visible light images can identify some cracks or stains, it is difficult to determine the light transmittance inside the shell and any abnormalities in the internal structure. Conversely, structural assessment based solely on near-infrared images often ignores the influence of appearance uniformity or morphological symmetry, causing quail eggs with abnormal appearances or deformations to be misclassified as normal. Furthermore, the lack of an effective multi-source information fusion mechanism and unified evaluation indicators makes it difficult to utilize detection results from different modalities in a coordinated manner, ultimately affecting the overall accuracy of detection and the scientific validity of quality grading. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a smart detection method and system for quail eggs based on multi-source data fusion, which solves the problem of the lack of multi-source information fusion and unified evaluation mechanism in existing technologies, leading to one-sided and inaccurate judgment of quail egg quality.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a smart detection method for quail eggs based on multi-source data fusion, comprising the following steps: acquiring multi-source detection data of quail eggs, including visible light detection images, near-infrared detection images, and geometric parameter data; extracting image texture features and spectral response features of the quail eggs to be detected based on the visible light detection images and near-infrared detection images, respectively; analyzing the image balance index and structural transmission index of the quail eggs to be detected based on the image texture features and spectral response features, respectively; analyzing the geometric morphology index of the quail eggs to be detected based on the morphological parameter data; analyzing the comprehensive state index of the quail eggs to be detected based on the image balance index, structural transmission index, and geometric morphology index; and determining the quality grade of the quail eggs to be detected based on the comprehensive state index.

[0006] Furthermore, the visible light detection image consists of several visible light pixels, and each visible light pixel corresponds to a visible light gray value. The near-infrared detection image consists of several near-infrared pixels, and each near-infrared pixel corresponds to a near-infrared reflection intensity value. The geometric parameter data includes the major axis length, minor axis length, and volume estimation value.

[0007] Furthermore, the image texture features include the image gray mean, image gray standard deviation, image gray offset, edge gradient amplitude, and bright area centroid offset distance. The spectral response features include the near-infrared reflectance intensity mean, near-infrared reflectance intensity variance, high reflectance pixel ratio, edge gray jump rate, and high-frequency region gray fluctuation amplitude.

[0008] Further, the specific steps for extracting the image texture features and spectral response features of the quail egg to be detected are as follows: read the visible light detection image and near-infrared detection image of the quail egg to be detected; input the visible light detection image and near-infrared detection image into the pre-trained image spectral model for extraction and analysis to obtain the image texture features and spectral response features of the quail egg to be detected, wherein the image spectral model includes a visible light extraction layer and a near-infrared extraction layer.

[0009] Further, the specific steps for obtaining the image texture features and spectral response features of the quail egg to be detected are as follows: In the visible light extraction layer of the image spectral model, grayscale and edge analysis are performed on the visible light detection image to obtain the image texture features of the quail egg to be detected; In the near-infrared extraction layer of the image spectral model, reflectance and frequency domain analysis are performed on the near-infrared detection image to obtain the spectral response features of the quail egg to be detected.

[0010] Further, the specific steps for analyzing the image balance index of the quail egg to be tested are as follows: read the image texture features of the quail egg to be tested and perform normalization processing; perform comprehensive analysis on the normalized image texture features to obtain the image balance index of the quail egg to be tested.

[0011] Further, the specific steps for analyzing the structural transmittance index of the quail egg to be tested are as follows: read the spectral response characteristics of the quail egg to be tested and perform normalization processing; perform comprehensive analysis on the normalized spectral response characteristics to obtain the structural transmittance index of the quail egg to be tested.

[0012] Further, the specific steps for analyzing the geometric morphology index of the quail egg to be tested are as follows: read the morphological parameter data of the quail egg to be tested and perform normalization processing; perform comprehensive analysis on the normalized morphological parameter data to obtain the geometric morphology index of the quail egg to be tested.

[0013] Furthermore, based on the comprehensive state index, the specific steps for determining the quality grade of the quail egg to be tested are as follows: the comprehensive state index of the quail egg to be tested is matched and analyzed with several preset state intervals, and each state interval corresponds to a quality grade; when the comprehensive state index is in a preset state interval, the quality grade corresponding to that state interval is determined as the quality grade of the quail egg to be tested.

[0014] A multi-source data fusion-based intelligent quail egg detection system includes: a multi-source data acquisition unit for acquiring multi-source detection data of quail eggs, including visible light detection images, near-infrared detection images, and geometric parameter data; an image extraction unit for extracting image texture features and spectral response features of the quail eggs to be detected based on the visible light detection images and near-infrared detection images, respectively; an image analysis unit for analyzing the image uniformity index and structural transmission index of the quail eggs to be detected based on the image texture features and spectral response features, respectively; a geometric analysis unit for analyzing the geometric morphology index of the quail eggs to be detected based on the morphological parameter data; a comprehensive analysis unit for analyzing the comprehensive state index of the quail eggs to be detected based on the image uniformity index, structural transmission index, and geometric morphology index; and a quality grade determination unit for determining the quality grade of the quail eggs to be detected based on the comprehensive state index.

[0015] The present invention has the following beneficial effects:

[0016] (1) The intelligent detection method for quail eggs based on multi-source data fusion analyzes the image texture features, spectral response features and morphological parameters by fusing visible light images, near-infrared images and geometric parameter data, extracts the image balance index, structural transmission index and geometric morphology index, and finally constructs a comprehensive state index, which can evaluate the comprehensive quality of the quail egg's exterior and interior. For example, a quail egg with an intact surface but extremely low near-infrared transmittance may be misjudged as normal in traditional image detection, but this invention will identify its potential abnormalities through the structural transmission index, thereby accurately classifying its quality.

[0017] (2) The intelligent detection method for quail eggs based on multi-source data fusion introduces parameters such as image texture features and edge gradient, near-infrared reflection and jump information, and high-frequency fluctuation amplitude during the detection process to describe the microscopic changes on the surface of quail eggs and the local anomalies in the internal structure. It can identify those problem areas that are difficult to detect with the naked eye and are easily ignored by traditional algorithms. For example, a quail egg with an intact shell but slight color deviation or local transmission abnormality at the edge is easily released by existing technology because the overall gray level is normal. However, this invention can capture such slight asymmetry or structural imbalance in a timely manner through detailed features such as the shift of the center of gravity of the bright area and the gray level fluctuation of the high-frequency area, and screen out risky eggs in advance, reducing the defective product outflow rate in downstream processing or sales. Thus, without increasing the complexity of the equipment, it significantly enhances the detection accuracy of hidden anomalies.

[0018] (3) The intelligent detection method for quail eggs based on multi-source data fusion sets up quality grade division intervals based on comprehensive state index, with each grade corresponding to a clear quality description, thereby realizing intelligent and standardized graded processing. This mechanism can be widely applied to the automatic sorting system of quail egg production enterprises, realizing effective decoupling between machine judgment and manual operation, avoiding quality fluctuation problems caused by subjective human judgment. This intelligent grade mechanism improves product circulation efficiency and also helps to unify the internal quality standards and sorting strategies of enterprises.

[0019] (4) The quail egg intelligent detection system based on multi-source data fusion is a modular intelligent detection system consisting of a multi-source data acquisition unit, an image extraction unit, an image analysis unit, a geometric analysis unit, a comprehensive analysis unit, and a quality grade determination unit. This system enables independent calling and decoupled operation of each functional module, making the system adaptable to different scales and application scenarios. For example, in a small sorting workshop, the core image analysis and quality determination modules can be retained and quickly operated with basic photography and weighing devices. In a large processing line, the system can be integrated into an automatic conveying device and combined with a multi-channel parallel processing framework to achieve real-time multi-egg detection and high-speed grading. The modular architecture also facilitates later upgrades and maintenance.

[0020] Of course, any product implementing this invention does not necessarily need to achieve all of the advantages described above at the same time. Attached Figure Description

[0021] Figure 1 is a flowchart of a smart detection method for quail eggs based on multi-source data fusion according to the present invention.

[0022] Figure 2 is a flowchart of the specific steps for analyzing the image balance index of the quail egg to be detected in the intelligent detection method for quail eggs based on multi-source data fusion of the present invention.

[0023] Figure 3 is a block diagram of a quail egg intelligent detection system based on multi-source data fusion according to the present invention. Detailed Implementation

[0024] Please refer to Figure 1. This embodiment of the invention provides a technical solution: an intelligent detection method for quail eggs based on multi-source data fusion, comprising the following steps: acquiring multi-source detection data of quail eggs, including visible light detection images, near-infrared detection images, and geometric parameter data; extracting image texture features and spectral response features of the quail egg to be detected based on the visible light detection images and near-infrared detection images, respectively; analyzing the image balance index and structural transmission index of the quail egg to be detected based on the image texture features and spectral response features, respectively; analyzing the geometric morphology index of the quail egg to be detected based on the morphological parameter data; analyzing the comprehensive state index of the quail egg to be detected based on the image balance index, structural transmission index, and geometric morphology index; and determining the quality grade of the quail egg to be detected based on the comprehensive state index.

[0025] The visible light detection image consists of several visible light pixels, and each visible light pixel corresponds to a visible light gray value. The near-infrared detection image consists of several near-infrared pixels, and each near-infrared pixel corresponds to a near-infrared reflection intensity value. The geometric parameter data includes the major axis length, minor axis length, and volume estimation value.

[0026] Visible light detection images can be acquired by visible light image acquisition devices (such as industrial cameras) under standard visible light illumination conditions.

[0027] Near-infrared detection images can be acquired by an image acquisition module equipped with a near-infrared band imaging sensor under a set wavelength (such as 850nm or 940nm).

[0028] The major axis length is used to characterize the maximum axial diameter of the target quail egg in the image outline, and can be obtained in the following way:

[0029] The visible light image of the target quail egg was obtained. After grayscale normalization and noise reduction, the edge pixel set was extracted using the Canny edge detection algorithm.

[0030] Subsequently, ellipse fitting is performed based on this set of pixels, and the fitting result includes parameters such as major axis, minor axis, center point and orientation angle;

[0031] The major axis length is the maximum diameter of the fitted ellipse along the principal axis direction, and the unit is pixels. If the image resolution is known, it can be converted to an actual length unit (such as millimeters). In this embodiment, the unit is millimeters.

[0032] The minor axis length represents the smallest axial diameter of the fitted ellipse perpendicular to the major axis, and can be obtained in the following way:

[0033] Based on the completed ellipse fitting, the minor axis length value is directly extracted from the fitting parameters.

[0034] This value corresponds to the maximum span of the target quail egg in its image outline along the vertical main axis;

[0035] Similarly, the unit can be converted from pixels to the actual length unit based on the image resolution; in this embodiment, the unit is millimeters.

[0036] The volume estimate is used to represent the approximate geometric volume of the target quail egg. It can be estimated based on the lengths of the major and minor axes of the fitted ellipse, as follows:

[0037] Divide the length of the major axis by 2 to get the semi-major axis value, and divide the length of the minor axis by 2 to get the semi-minor axis value;

[0038] Subsequently, the volume was calculated using an ellipsoidal approximation model, which involved multiplying the semi-major axis by the two semi-minor axes to obtain a three-dimensional product, and then multiplying it by 4 / 3 and pi (approximately 3.1416).

[0039] The calculation formula can be described as: Volume estimate = (4 / 3) × π × semi-major axis × semi-minor axis × semi-minor axis;

[0040] The final volume estimate is in cubic millimeters.

[0041] The specific formula for calculating the comprehensive condition index of the quail eggs to be tested is as follows:

[0042] ;in, , , , The indices are, in order: comprehensive state index, image balance index, structural transmission index, and geometric morphology index of the quail egg to be tested. , , The values ​​are, in order, the image adjustment coefficient, the structural adjustment coefficient, and the geometric adjustment coefficient stored in the database, and in this embodiment, they are respectively set to 1.15, 1.30, and 0.85.

[0043] The image adjustment coefficient ranges from [0.5, 2.0].

[0044] The range of the structural adjustment coefficient is [0.5, 2.0].

[0045] The range of the geometric adjustment coefficient is: [0.5, 2.0].

[0046] Specifically, image texture features include image gray mean, image gray standard deviation, image gray offset, edge gradient amplitude, and bright area centroid offset distance; spectral response features include near-infrared reflectance intensity mean, near-infrared reflectance intensity variance, high reflectance pixel ratio, edge gray jump rate, and high-frequency region gray fluctuation amplitude.

[0047] The specific steps for extracting the image texture features and spectral response features of the quail egg to be detected are as follows: Read the visible light detection image and near-infrared detection image of the quail egg to be detected; input the visible light detection image and near-infrared detection image into the pre-trained image spectral model for extraction and analysis to obtain the image texture features and spectral response features of the quail egg to be detected, wherein the image spectral model includes a visible light extraction layer and a near-infrared extraction layer.

[0048] The pre-training steps for the image spectral model are as follows:

[0049] First, a large number of visible light and near-infrared images of quail eggs with known quality were collected to construct a training sample set containing different surface textures and internal spectral features. Then, feature annotation was performed on each set of images, and the corresponding image texture feature values ​​(such as the average gray level and edge gradient magnitude) and spectral response feature values ​​(such as the average near-infrared reflectance and edge gray level jump rate) were recorded to form structured annotation data.

[0050] In terms of model structure, the image spectral model includes a visible light extraction layer and a near-infrared extraction layer. The visible light extraction layer performs multi-scale gray-level statistics and edge feature extraction on the visible light image through a gray-level mapping module and an edge coding module. The near-infrared extraction layer performs pixel energy statistics and frequency domain perturbation analysis on the near-infrared image through a reflection mapping module and a frequency domain deconstruction module. Both extraction layers output structured feature vectors, which are used to fit the true labels in the training set.

[0051] During training, the input image must be of the same size and have the same number of channels. The training uses the mean squared error loss function and the Adam optimizer for gradient updates. The number of training epochs, learning rate, and batch size are configured according to the actual data size, such as a learning rate of 0.001, a batch size of 32, and 100 training epochs.

[0052] Finally, by comparing the extraction errors of the training set and the validation set, it is ensured that the various image texture features and spectral response features output by the model have a high degree of fit, and the average feature error is controlled within a preset range (such as within 5%), thereby completing the pre-training process of the image spectral model.

[0053] The specific steps for obtaining the image texture features and spectral response features of the quail egg to be detected are as follows: In the visible light extraction layer of the image spectral model, grayscale and edge analysis are performed on the visible light detection image to obtain the image texture features of the quail egg to be detected, specifically:

[0054] For the image grayscale mean:

[0055] Read the visible light grayscale values ​​of all visible light pixels in the visible light detection image;

[0056] Calculate the arithmetic mean of all gray values ​​to obtain the image gray mean.

[0057] For the standard deviation of image gray levels:

[0058] Calculate the square of the difference between the gray value of each visible light pixel and the mean gray value of the image;

[0059] The standard deviation of the image grayscale is obtained by averaging all the squared differences and then taking the square root.

[0060] For image grayscale offset values:

[0061] The difference between the average grayscale value of the image and the set standard reference grayscale value is calculated and used as the image grayscale offset value.

[0062] For edge gradient magnitude:

[0063] Edge detection processing is performed on the visible light detection image, as follows:

[0064] In the horizontal direction, the horizontal grayscale variation of each pixel is extracted using a convolution kernel [-1,0,1] or the Sobel operator;

[0065] In the vertical direction, the vertical grayscale variation of each pixel is extracted using a convolution kernel [1,2,1]^T or the Sobel operator;

[0066] Edge pixels are defined as pixels whose horizontal or vertical variation value exceeds a set threshold.

[0067] Calculate the gradient magnitude of each edge pixel:

[0068] For each edge pixel, its gradient magnitude is the square root of the sum of the square of the horizontal grayscale change and the square of the vertical grayscale change.

[0069] Take the arithmetic mean of the gradient magnitudes of all edge pixels as the edge gradient magnitude.

[0070] For the centroid offset distance value of the bright area:

[0071] Extract the set of bright area pixels in the visible light detection image whose grayscale value is higher than a set bright area threshold;

[0072] Calculate the arithmetic mean of the x and y coordinates of all bright area pixels, and use it as the centroid coordinates of the bright area;

[0073] Next, calculate the Euclidean distance between the centroid coordinates of the bright area and the image center coordinates, and use it as the centroid offset distance value of the bright area;

[0074] In the near-infrared extraction layer of the image spectral model, reflectance and frequency domain analysis are performed on the near-infrared detection image to obtain the spectral response characteristics of the quail egg to be detected, specifically:

[0075] For the mean near-infrared reflectance intensity:

[0076] Read the near-infrared reflectance intensity values ​​of all near-infrared pixels in the near-infrared detection image;

[0077] Calculate the arithmetic mean of all reflection intensity values ​​to obtain the mean near-infrared reflection intensity.

[0078] For the near-infrared reflectance intensity variance:

[0079] Calculate the square of the difference between the reflection intensity value of each near-infrared pixel and the mean near-infrared reflection intensity;

[0080] The arithmetic mean of all squared differences is taken as the variance of near-infrared reflectance intensity.

[0081] For the proportion of highly reflective pixels:

[0082] Set a high reflectivity threshold;

[0083] Count the number of pixels in the near-infrared detection image whose reflection intensity value is higher than this threshold;

[0084] Divide the number of highly reflective pixels by the total number of pixels in the image to obtain the percentage of highly reflective pixels.

[0085] For edge grayscale jump rate values:

[0086] Perform edge detection as follows:

[0087] The Sobel operator is used to calculate the grayscale changes of each pixel in the horizontal and vertical directions;

[0088] Pixels whose change value exceeds a set threshold are marked as edge pixels;

[0089] Extract the maximum and minimum gray values ​​of each edge pixel in its 3×3 neighborhood and calculate their difference;

[0090] The average of all differences is used to obtain the edge grayscale jump amplitude value;

[0091] Divide this value by the average near-infrared reflectance intensity;

[0092] Obtain the edge grayscale jump rate value.

[0093] For the grayscale fluctuation amplitude in the high-frequency region:

[0094] Perform a frequency domain transformation (such as a fast Fourier transform) on the near-infrared detection image, specifically:

[0095] Perform a two-dimensional fast Fourier transform (FFT) on the near-infrared detection image to obtain a frequency domain image;

[0096] Set a high-frequency threshold, and select the region corresponding to the frequency domain coefficients with frequencies greater than the threshold as the high-frequency region;

[0097] Extract high-frequency regions from the frequency domain image whose frequencies are higher than a set high-frequency threshold;

[0098] In the high-frequency region, the difference between the maximum and minimum grayscale values ​​of all pixels is calculated and used as the grayscale fluctuation amplitude in the high-frequency region.

[0099] In this implementation scheme, by introducing visible light texture features such as image grayscale, edge gradient, and bright area centroid, as well as spectral response features such as near-infrared reflectance mean, high reflectance pixel ratio, and grayscale jump rate, this method can comprehensively evaluate the quality status of quail eggs from multiple perspectives, including image brightness distribution, edge contour clarity, and internal transmission details. Compared with traditional methods that only look at overall brightness or external cracks, these features can more meticulously capture subtle anomalies such as slight shell thinning, local color shift, and structural asymmetry. For example, although an egg may not have visible cracks on its surface, if its near-infrared image shows a large variance in reflectance intensity and violent grayscale fluctuations in high-frequency areas, it indicates an uneven shell structure and potential quality problems. If the centroid of the bright area is significantly shifted, it may also mean that the eggshell is subjected to uneven stress or deformation, thus helping the system to identify risky samples earlier and reduce the false positive rate.

[0100] Specifically, as shown in Figure 2, the specific steps for analyzing the image balance index of the quail egg to be tested are as follows: read the image texture features of the quail egg to be tested and perform normalization processing; perform comprehensive analysis on the normalized image texture features to obtain the image balance index of the quail egg to be tested.

[0101] The specific formula for calculating the image balance index of the quail egg to be detected is as follows:

[0102] ;in, , , , , , The following parameters are listed in order: image balance index, image grayscale mean, image grayscale standard deviation, image grayscale offset, edge gradient magnitude, and bright area centroid offset distance of the quail egg to be tested. , , , , The values ​​are, in order, the grayscale adjustment coefficient, contrast adjustment coefficient, grayscale offset adjustment coefficient, edge disturbance adjustment coefficient, and bright area offset adjustment coefficient stored in the database, and in this embodiment, they are taken as 1.20, 1.10, 0.85, 1.05, and 1.25 respectively.

[0103] The grayscale adjustment coefficient ranges from [0.5, 2.0].

[0104] The contrast adjustment coefficient ranges from [0.5, 2.0].

[0105] The grayscale offset adjustment coefficient has a range of values: [0.5, 2.0].

[0106] The range of the edge disturbance adjustment coefficient is [0.5, 1.5].

[0107] The value range of the bright area offset adjustment coefficient is [0.5, 2.0].

[0108] In this implementation scheme, by constructing an image balance index, the image quality and imaging uniformity of the quail egg surface can be more comprehensively reflected. Unlike the traditional processing method that only looks at grayscale or edge intensity, this method not only considers the overall brightness level and contrast of the image, but also takes into account information such as local disturbances and structural shifts. It is especially suitable for dealing with complex situations such as uneven image brightness, spot shifts, and blurred contours. By setting a reasonable adjustment coefficient range, the system can flexibly adapt to the image characteristics of different batches and different lighting environments, and improve the stability of the algorithm in identifying image anomalies. For example, although some egg surfaces have no obvious dirt, the distribution of bright areas is abnormal or the edge changes drastically. Traditional algorithms may ignore this, but the image balance index can accurately reflect the poor overall image condition, thereby effectively avoiding misjudgment.

[0109] Specifically, the steps for analyzing the structural transmittance index of the quail egg to be tested are as follows: read the spectral response characteristics of the quail egg to be tested and normalize them; perform a comprehensive analysis on the normalized spectral response characteristics to obtain the structural transmittance index of the quail egg to be tested.

[0110] The specific formula for calculating the structural transmittance index of the quail egg to be tested is as follows:

[0111] ;in, , , , , , The following parameters are, in order: structural transmittance index, mean infrared reflectance intensity, variance of near-infrared reflectance intensity, proportion of high reflectance pixels, edge grayscale jump rate, and grayscale fluctuation amplitude in high-frequency regions of the quail egg to be tested. , , The values ​​are, in order, the reflection fusion adjustment coefficient, the high reflection ratio adjustment coefficient, and the edge height composite adjustment coefficient stored in the database, and in this embodiment, they are respectively taken as 0.55, 0.70, and 1.10.

[0112] The grayscale adjustment coefficient ranges from [0.1, 1.0].

[0113] The contrast adjustment coefficient ranges from [0.2, 1.2].

[0114] The grayscale offset adjustment coefficient has a range of values: [0.3, 1.5].

[0115] In this implementation scheme, the structural transmittance index is used to fuse multiple spectral features in near-infrared images, effectively reflecting the transmittance of infrared light by quail eggshells and thus determining whether there are any abnormalities in their internal structure. Unlike traditional methods that rely solely on surface image recognition, this method introduces parameters such as the mean of infrared reflection intensity, reflection variance, and the proportion of high-reflectivity pixels. This allows for a deeper understanding of transmittance anomalies caused by issues such as uneven density, thickness, or microcracks within the eggshell. Furthermore, edge grayscale jump rate and high-frequency region fluctuation amplitude further supplement the dimensions of boundary clarity and detail stability in the image, improving the overall accuracy of the judgment. Through normalization and preset adjustment coefficients, this method can effectively suppress interference caused by inconsistencies in the dimensions of features, avoiding the risk of misjudgment where a single outlier masks the overall performance. For example, for a quail egg with an intact exterior but high-density patches inside, its structural transmittance index will be significantly lower, and the system can promptly mark it as a potential anomaly, improving the ability of the grading process to perceive internal defects and thus compensating for the shortcomings of traditional methods in identifying internal quality.

[0116] Specifically, the steps for analyzing the geometric morphology index of the quail egg to be tested are as follows: read the morphological parameter data of the quail egg to be tested and normalize it; perform a comprehensive analysis on the normalized morphological parameter data to obtain the geometric morphology index of the quail egg to be tested.

[0117] The specific formula for calculating the geometric morphology index of the quail egg to be tested is as follows:

[0118] ;in, , , , The following are the geometric morphology indexes, estimated volume, major axis length, and minor axis length of the quail egg to be tested, in that order. , The values ​​are the volume adjustment coefficient and the morphology ratio adjustment coefficient stored in the database, respectively, and in this embodiment, they are 1.05 and 1.15, respectively.

[0119] The volume adjustment coefficient ranges from [0.5, 2.0].

[0120] The range of the morphology ratio adjustment coefficient is [0.5, 2.0].

[0121] In this implementation plan, the geometric morphology index is used to integrate the volume and major-minor axis ratio of quail eggs for comprehensive evaluation of the overall symmetry and structural rationality of the egg's appearance. This addresses the problem of traditional testing methods that rely solely on images without considering morphology. By introducing estimated volume values ​​and major-minor axis length data, it can accurately reflect whether the egg is excessively long and thin, flat, or deformed, avoiding the misjudgment of eggs with smooth appearances but abnormal shapes as high-quality eggs. Furthermore, normalization ensures a consistent numerical dimension for different morphological parameters, preventing unreasonable deviations in the final judgment due to absolute differences in volume or length values. The adjustment coefficient also provides a degree of flexibility, allowing for adjustments to the influence weight of different parameters based on actual quality control needs. For example, a quail egg with a flat, round shape and moderate size will have a significantly higher geometric morphology index than a sample with a long, thin shape or asymmetrical ends, contributing to the achievement of graded management goals for appearance stability and specification uniformity.

[0122] Specifically, the steps for determining the quality grade of the quail egg to be tested based on the comprehensive state index are as follows: The comprehensive state index of the quail egg to be tested is matched and analyzed with several preset state intervals, with each state interval corresponding to a quality grade; when the comprehensive state index falls within a preset state interval, the quality grade corresponding to that state interval is determined as the quality grade of the quail egg to be tested, including but not limited to the following examples:

[0123] When the comprehensive condition index is in the range of [0.00, 0.30], the corresponding quality grade is level 5 (unqualified), which means that the quail egg has serious abnormalities in terms of image texture, structural transmission and geometric shape. It may have problems such as cracks, significant deformation or abnormal reflection of contents, and does not meet the factory requirements.

[0124] When the comprehensive condition index is in the range of (0.30, 0.55), the corresponding quality grade is level four (poor), which means that the quail egg has poor near-infrared transmittance or low image uniformity, and may have problems such as slight structural damage or cloudy contents inside the shell.

[0125] When the comprehensive condition index is in the range of (0.55, 0.75), the corresponding quality grade is level three (general), which means that the overall condition of the quail egg is general. Although no obvious abnormalities are found, there are slight deviations in image texture or reflection distribution.

[0126] When the comprehensive condition index is in the range of (0.75, 0.90), the corresponding quality grade is level two (good), which means that the quail egg has a clear image, stable structure, and no obvious spectral abnormalities, and is suitable for standard commodity channels such as supermarket retail.

[0127] When the comprehensive quality index is in the range of (0.90, 1.05), the corresponding quality grade is Grade 1 (superior), which means that the quail egg has an intact surface structure, good internal transmission characteristics, symmetrical shape, and excellent overall quality.

[0128] In this implementation plan, the automatic grading of quail egg quality is achieved by corresponding the comprehensive state index with preset state intervals. This effectively avoids the subjectivity and inconsistency problems caused by traditional reliance on manual experience. Each grade interval is set based on the fusion analysis results of multi-dimensional indicators such as image texture, structural transmission, and geometric morphology, which can comprehensively reflect the overall level of the egg's appearance and internal condition. This method can not only accurately identify unqualified products with serious problems such as cracks, deformation, or abnormal reflection, but also reasonably classify samples in borderline states, reducing missed detections and misjudgments. For example, eggs with clear images but insufficient transmission will no longer be misjudged as superior simply because of good performance in a single dimension; conversely, eggs with average morphology but no obvious abnormalities will not be over-downgraded. This interval-based grading method makes the quality grading results more stable, transparent, and practical, which is conducive to promoting the standardization of quality control at the production end.

[0129] Please refer to Figure 3. This embodiment of the invention provides a technical solution: an intelligent quail egg detection system based on multi-source data fusion, comprising: a multi-source data acquisition unit for acquiring multi-source detection data of quail eggs, including visible light detection images, near-infrared detection images, and geometric parameter data; an image extraction unit for extracting image texture features and spectral response features of the quail egg to be detected based on the visible light detection images and near-infrared detection images, respectively; an image analysis unit for analyzing the image balance index and structural transmission index of the quail egg to be detected based on the image texture features and spectral response features, respectively; a geometric analysis unit for analyzing the geometric morphology index of the quail egg to be detected based on the morphological parameter data; a comprehensive analysis unit for analyzing the comprehensive state index of the quail egg to be detected based on the image balance index, structural transmission index, and geometric morphology index; and a quality grade determination unit for determining the quality grade of the quail egg to be detected based on the comprehensive state index.

[0130] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention.

[0131] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A smart detection method for quail eggs based on multi-source data fusion, characterized in that, Includes the following steps: Multi-source detection data of quail eggs is acquired, including visible light detection images, near-infrared detection images, and geometric parameter data. Based on the visible light and near-infrared detection images, image texture features and spectral response features of the quail eggs to be detected are extracted, specifically: the visible light and near-infrared detection images of the quail eggs to be detected are read; the visible light and near-infrared detection images are input into a pre-trained image spectral model for extraction and analysis. The image spectral model includes a visible light extraction layer and a near-infrared extraction layer to obtain the image texture features and spectral response features of the quail eggs to be detected. Specifically: in the visible light extraction layer of the image spectral model, grayscale and edge analysis are performed on the visible light detection images to obtain the image texture features of the quail eggs to be detected, including the image grayscale mean, image grayscale standard deviation, image grayscale offset value, edge gradient magnitude, and bright area centroid offset distance value. In the near-infrared extraction layer of the image spectral model, reflectance and frequency domain analysis are performed on the near-infrared detection image to obtain the spectral response characteristics of the quail egg to be detected, including the mean value of near-infrared reflectance intensity, the variance of near-infrared reflectance intensity, the proportion of high reflectance pixels, the edge gray-level jump rate value, and the gray-level fluctuation amplitude in high-frequency regions. Based on the image texture features and spectral response features, the image balance index and structural transmission index of the quail egg to be detected are analyzed respectively. Based on morphological parameter data, the geometric morphology index of the quail egg to be tested is analyzed; based on image balance index, structural transmission index, and geometric morphology index, the comprehensive state index of the quail egg to be tested is analyzed, specifically as follows: ;in, 、 、 、 The indices are, in order: comprehensive state index, image balance index, structural transmission index, and geometric morphology index of the quail egg to be tested. 、 、 These are, in order, the image adjustment coefficients, structural adjustment coefficients, and geometric adjustment coefficients stored in the database; Based on the comprehensive state index, the quality grade of the quail egg to be tested is determined. Specifically, the comprehensive state index of the quail egg to be tested is matched and analyzed with several preset state intervals, and each state interval corresponds to a quality grade. When the comprehensive state index is in a preset state interval, the quality grade corresponding to that state interval is determined as the quality grade of the quail egg to be tested.

2. The intelligent quail egg detection method based on multi-source data fusion according to claim 1, characterized in that, The visible light detection image consists of several visible light pixels, and each visible light pixel corresponds to a visible light gray value. The near-infrared detection image consists of several near-infrared pixels, and each near-infrared pixel corresponds to a near-infrared reflection intensity value. The geometric parameter data includes the major axis length, minor axis length, and volume estimation value.

3. The intelligent detection method for quail eggs based on multi-source data fusion according to claim 1, characterized in that, The specific steps for analyzing the image balance index of the quail egg to be tested are as follows: read the image texture features of the quail egg to be tested and perform normalization processing; perform comprehensive analysis on the normalized image texture features to obtain the image balance index of the quail egg to be tested.

4. The intelligent detection method for quail eggs based on multi-source data fusion according to claim 1, characterized in that, The specific steps for analyzing the structural transmittance index of the quail egg to be tested are as follows: read the spectral response characteristics of the quail egg to be tested and normalize them; perform a comprehensive analysis on the normalized spectral response characteristics to obtain the structural transmittance index of the quail egg to be tested.

5. The intelligent detection method for quail eggs based on multi-source data fusion according to claim 1, characterized in that, The specific steps for analyzing the geometric morphology index of the quail egg to be tested are as follows: read the morphological parameter data of the quail egg to be tested and normalize it; perform comprehensive analysis on the normalized morphological parameter data to obtain the geometric morphology index of the quail egg to be tested.

6. A quail egg intelligent detection system based on multi-source data fusion, employing the quail egg intelligent detection method based on multi-source data fusion as described in any one of claims 1-5, characterized in that, include: The multi-source data acquisition unit is used to acquire multi-source detection data of quail eggs, including visible light detection images, near-infrared detection images, and geometric parameter data; The image extraction unit is used to extract the image texture features and spectral response features of the quail egg to be detected based on the visible light detection image and the near-infrared detection image, respectively. The image analysis unit is used to analyze the image uniformity index and structural transmission index of the quail egg to be tested based on image texture features and spectral response features, respectively; the geometric analysis unit is used to analyze the geometric morphology index of the quail egg to be tested based on morphological parameter data. The comprehensive analysis unit is used to analyze the comprehensive state index of the quail eggs to be tested based on the image balance index, structural transmission index, and geometric morphology index; the quality grade determination unit is used to determine the quality grade of the quail eggs to be tested based on the comprehensive state index.

Images