Fruit damage assessment method and device in camellia oleifera fruit picking process

Abstract

The invention discloses a fruit damage assessment method and device in a camellia oleifera fruit picking process, and relates to the technical field of damage assessment. Comprising the following steps: acquiring a plurality of camellia oleifera fruit samples and physiological feature parameters thereof, acquiring spectral image data of fruits, and preprocessing the spectral image data to generate a training sample data set; and establishing a convolutional neural network model based on the data set, inputting spectral image data, and performing training by taking the physiological feature parameters as labels to obtain a physiological feature parameter prediction model. In the picking process, spectrum and visible light true color images of fruits to be detected are obtained, and physiological and appearance characteristic parameters are extracted. And calculating a physiological injury influence coefficient and an external injury influence coefficient according to the predicted value, comprehensively obtaining a fruit injury evaluation index, comparing the fruit injury evaluation index with a dynamically corrected injury degree judgment threshold, and sending out a corresponding injury state judgment result. Damage evaluation of camellia oleifera fruits can be realized, and the picking quality is improved.

Description

Technical Field

[0001] The invention relates to the technical field of damage assessment, and in particular to a method and a device for assessing damage to tea fruits during the picking process. Background Art

[0002] During the picking and transportation of tea fruit, fruit damage is an important issue that directly affects the quality and economic value of the fruit. Tea fruit is an important oil crop, and its fruit is rich in oil, nutrients and active substances. With the increasing market demand for high-quality tea fruit, how to effectively evaluate and control fruit damage during the picking process has become a technical problem that needs to be solved urgently in the industry. At present, traditional fruit damage assessment methods mainly rely on manual visual inspection and mechanical measurement, which is not only time-consuming and labor-intensive, but also has poor accuracy and is prone to subjective errors. It is unable to timely and comprehensively reflect the true state of the fruit.

[0003] In existing research, although multispectral imaging technology has made some progress in the determination of plant physiological characteristics, its application in the assessment of oil-tea fruit damage is still in the exploratory stage. Multispectral imaging can capture the spectral information of the fruit and reflect its internal physiological characteristics, such as changes in moisture, acids and other components, which are closely related to the health of the fruit. However, how to establish an effective mapping relationship between these spectral features and fruit damage, and then form an efficient and accurate assessment model, is still a technical challenge. In addition, existing damage assessment methods often ignore the impact of fruit appearance characteristics on its quality, resulting in one-sidedness and limitations in the assessment results.

[0004] In the prior art, the publication number CN112329296A discloses a method, device, computer equipment and storage medium for calculating mechanical damage of fruits, and the method mainly includes: obtaining three-dimensional data and regional measurement data of fruits with the same middle diameter, wherein the three-dimensional data includes the height, upper diameter, middle diameter and lower diameter of the fruits; establishing an associated dimension chain based on the middle diameter according to the three-dimensional data and regional measurement data, wherein the other data in the associated dimension chain are proportional to the middle diameter; constructing a fruit model according to the associated dimension chain; and meshing the fruit model to construct a finite element model of a fruit harvesting machine; setting the motion parameters of the fruit and the fruit harvesting machine, and obtaining the force and deformation cloud map of the fruit to determine the damage ratio of the fruit. However, this method only considers the external damage of the fruit, and does not consider the physiological influencing parameters of the fruit, and the possible internal damage of the fruit, such as rot, deterioration, etc., thereby reducing the accuracy and effectiveness of the damage assessment.

[0005] The above information disclosed in this Background section is only for enhancement of understanding of the background of the present disclosure and therefore it may contain information that does not constitute the prior art that is already known to one of ordinary skill in the art. Summary of the invention

[0006] The purpose of the present invention is to provide a method and device for assessing fruit damage during the picking process of tea-oil fruits, so as to solve the problems raised in the above-mentioned background technology.

[0007] To achieve the above object, the present invention provides the following technical solutions: A method for assessing fruit damage during tea-oil fruit picking, comprising the following specific steps: Acquire several camellia fruit samples with known physiological characteristic parameters of camellia fruit, acquire spectral image data of the camellia fruit samples by using a multispectral imaging device, pre-process the acquired spectral image data to obtain sample spectral image data, and map the sample spectral image data to the corresponding physiological characteristic parameters of camellia fruit one by one to generate a training sample data set, wherein the physiological characteristic parameters include water content, acetic acid content, ethanol content and butyric acid content; Based on the data in the training sample data set, a convolutional neural network model is established, the sample spectral image data in the training sample data set is used as the input of the convolutional neural network model, and the physiological characteristic parameters of tea fruit corresponding to the sample spectral image are used as labels to train the convolutional neural network model to obtain a physiological characteristic parameter prediction model; During the picking process, a spectral image of the oil-tea camellia fruit to be detected is obtained, and at the same time, a visible light true color image of the oil-tea camellia fruit to be detected is collected, and after preprocessing the spectral image of the oil-tea camellia fruit to be detected, the image is input into the trained physiological characteristic parameter prediction model to obtain the predicted value of the physiological characteristic parameter of the oil-tea camellia fruit to be detected, and the appearance characteristic parameters of the oil-tea camellia fruit to be detected are extracted according to the visible light true color image of the oil-tea camellia fruit to be detected; The physiological damage influence coefficient of the camellia fruit is calculated based on the predicted value of the physiological characteristic parameter of the camellia fruit to be tested, and the external damage influence coefficient of the camellia fruit is calculated based on the appearance characteristic parameter of the camellia fruit to be tested and the surface roughness of the camellia fruit to be tested, wherein the appearance characteristic parameter of the camellia fruit to be tested includes the contour circumference, contour area and color contrast of the camellia fruit; Based on the obtained physiological damage influence coefficient and external damage influence coefficient of tea oil fruit, the fruit damage assessment index is calculated comprehensively, and the generated fruit damage assessment index is compared with the damage degree judgment threshold. According to different comparison results, the corresponding damage status judgment result is issued, among which the damage degree judgment threshold is dynamically corrected according to the environmental humidity of the place where the tea oil fruits are picked.

[0008] Further, spectral image data of the oil-tea camellia fruit sample is obtained by a multispectral imaging device, wherein the wavelength range of the multispectral imaging device is 1400 nanometers to 2500 nanometers; Acquiring spectral image data of the camellia fruit sample by a multispectral imaging device, and performing image preprocessing on the spectral image data of the camellia fruit sample, wherein the preprocessing includes image enhancement and denoising preprocessing, wherein a wavelet transform denoising method is used to perform denoising on each spectral image of the camellia fruit sample, and a bilateral filter is used to perform image enhancement preprocessing on each spectral image of the camellia fruit sample; The method for generating the training sample data set is as follows: the sample spectral image data is mapped one by one with the physiological characteristic parameters of the corresponding tea fruit to form a corresponding grid, and the formed grid is recorded as the training sample data set.

[0009] Furthermore, based on the data in the training sample data set, a convolutional neural network model is established, and based on the convolutional neural network, a physiological characteristic parameter prediction model is established, wherein the convolutional neural network is composed of an input layer, a convolutional layer, a pooling layer, a fully connected layer and an output layer, wherein the activation function in the convolutional layer is function, The specific expression of the function is: ; in, Indicates The corresponding convolutional layers, It means the The corresponding convolutional layer The first training sample image eigenvalues, where is the index of the convolutional layer, is the index of the training sample image, is the index of the feature value in the training sample image; For the fully connected layer, the number of neurons in the fully connected layer is set to 32, the initial neural network learning rate is set to 0.001, and the number of training rounds is 200; The trained physiological characteristic parameter prediction model takes as input the spectral image data of the tea fruit, and outputs the predicted values ​​of the physiological characteristic parameters of the corresponding tea fruit.

[0010] Furthermore, according to the predicted value of the physiological characteristic parameter of the oil-tea camellia fruit to be tested, the influence coefficient of physiological damage of the oil-tea camellia fruit is calculated and generated, wherein the formula for calculating the influence coefficient of physiological damage of the oil-tea camellia fruit is: ; In the formula, is the physiological damage influence coefficient of Camellia oleifera fruit, is the predicted value of water content of the oil-tea camellia fruit to be tested, is the predicted value of ethanol content in the oil-tea camellia fruit to be tested, is the predicted value of acetic acid content in the oil-tea camellia fruit to be tested, It is the predicted value of butyric acid content in the Camellia oleifera fruit to be tested.

[0011] Furthermore, based on the appearance characteristic parameters of the oil-tea camellia fruit to be tested and the surface roughness of the oil-tea camellia fruit to be tested, the influence coefficient of external damage to the oil-tea camellia fruit is calculated and generated, wherein the formula for calculating the influence coefficient of external damage to the oil-tea camellia fruit is: ; In the formula, is the influence coefficient of external damage to camellia oleifera fruit, is the color contrast of the tea fruit to be tested, is the roughness of the skin of the Camellia oleifera fruit to be tested, is the roundness of the outline of the oil-tea camellia fruit to be tested; The circularity of the contour of the oil-tea camellia fruit to be detected is calculated by the appearance characteristic parameters of the oil-tea camellia fruit to be detected, and the specific formula for calculating the circularity of the contour of the oil-tea camellia fruit to be detected is: ; In the formula, is the outline area of ​​the tea fruit to be tested, is the circumference of the camellia oleifera fruit to be tested.

[0012] Furthermore, the epidermal roughness of the oil-tea camellia fruit to be tested is The specific acquisition method is: the skin roughness of the oil-tea fruit to be tested is detected by laser scanning. During the scanning process, the laser will form point cloud data on the surface of the fruit to record the shape and microscopic features of the surface.

[0013] Furthermore, based on the obtained physiological damage influence coefficient and external damage influence coefficient of Camellia oleifera fruit, the fruit damage assessment index is comprehensively calculated, wherein the formula for calculating the fruit damage assessment index is: ; In the formula, is the fruit damage assessment index, and are the weight coefficients of the influence coefficient of physiological damage of Camellia oleifera fruit and the influence coefficient of external damage of Camellia oleifera fruit, respectively. and and All are greater than 0; The generated fruit damage assessment index is compared with the damage degree judgment threshold, and the corresponding damage status judgment result is issued according to different comparison results. The specific judgment logic is as follows: when When the damage is detected, the tea fruit to be tested is judged to be first-grade damaged, indicating that the tea fruit to be tested cannot meet the needs of eating and processing; when When the damage is detected, the oil-tea camellia fruit to be tested is judged to be of the second-level damage, which means that the oil-tea camellia fruit to be tested cannot be sold directly and is used for secondary processing and sales; when When the oil-tea camellia fruit to be tested is judged to be level three damaged, it means that the oil-tea camellia fruit to be tested is intact and can be sold directly; in is the damage degree judgment threshold, which is obtained by dynamically correcting the ambient humidity at the location where the oil-tea fruit is picked. The calculation is based on the formula: ; In the formula, is the initial value of the damage degree judgment threshold, The environmental humidity of the tea fruit picking area. For reference humidity.

[0014] The present invention also provides a device for assessing damage to fruits during the picking process of oil-tea fruit, and the device for assessing damage to fruits during the picking process of oil-tea fruit is used to execute the above-mentioned method for assessing damage to fruits during the picking process of oil-tea fruit, comprising: A training data acquisition module is used to obtain several camellia fruit samples with known physiological characteristic parameters of camellia fruit, obtain spectral image data of the camellia fruit samples through a multispectral imaging device, pre-process the obtained spectral image data to obtain sample spectral image data, and map the sample spectral image data to the corresponding physiological characteristic parameters of the camellia fruit one by one to generate a training sample data set, wherein the physiological characteristic parameters include water content, acetic acid content, ethanol content and butyric acid content; The prediction model training module is used to establish a convolutional neural network model based on the data in the training sample data set, use the sample spectral image data in the training sample data set as the input of the convolutional neural network model, and use the physiological characteristic parameters of the oil-tea camellia fruit corresponding to the sample spectral image as labels to train the convolutional neural network model to obtain a physiological characteristic parameter prediction model; An appearance feature extraction module is used to obtain a spectral image of the oil-tea camellia fruit to be detected during the picking process, and simultaneously collect a visible light true color image of the oil-tea camellia fruit to be detected, and after pre-processing the spectral image of the oil-tea camellia fruit to be detected, input it into a trained physiological characteristic parameter prediction model to obtain a predicted value of the physiological characteristic parameter of the oil-tea camellia fruit to be detected, and extract the appearance feature parameters of the oil-tea camellia fruit to be detected based on the visible light true color image of the oil-tea camellia fruit to be detected; A damage factor analysis module is used to calculate and generate a camellia fruit physiological damage influence coefficient based on the predicted value of the camellia fruit physiological characteristic parameter to be detected, and to calculate and generate a camellia fruit external damage influence coefficient based on the appearance characteristic parameters of the camellia fruit to be detected and the epidermis roughness of the camellia fruit to be detected, wherein the appearance characteristic parameters of the camellia fruit to be detected include the camellia fruit contour circumference, contour area and color contrast; The comprehensive damage assessment module is used to comprehensively calculate the fruit damage assessment index based on the obtained physiological damage influence coefficient and external damage influence coefficient of the oil-tea fruit, compare the generated fruit damage assessment index with the damage degree judgment threshold, and issue corresponding damage status judgment results based on different comparison results, among which the damage degree judgment threshold is dynamically corrected by the environmental humidity of the place where the oil-tea fruit is picked.

[0015] Compared with the prior art, the present invention has the following beneficial effects: First, by obtaining tea fruit samples with known physiological characteristic parameters, the spectral image data of the fruit is accurately measured using multispectral imaging technology, and preprocessed to establish a mapping relationship between the sample spectrum and the physiological characteristic parameters. The deep learning model is used to extract the intrinsic connection between the physiological characteristics of the fruit and the degree of damage, paving the way for the following specific processing. Secondly, the convolutional neural network model established based on the training sample data set has efficient learning ability and strong prediction ability, and can quickly and accurately predict the physiological characteristic parameters of the tea fruit to be tested. This prediction is not limited to the basic physiological parameters of the fruit, but can also further reduce the error based on the appearance characteristics, thereby realizing a comprehensive assessment of the fruit damage. Combined with the appearance characteristic parameters extracted from the visible light true color image, such as contour perimeter, contour area and color contrast, the model can more comprehensively reflect the health status of the fruit. Finally, through the comprehensive calculation of the physiological damage influence coefficient and the external damage influence coefficient, the fruit damage assessment index is obtained, and it is compared with the dynamically corrected damage degree judgment threshold, which can accurately judge the damage status of the fruit in real time. It improves the management efficiency of the oil-tea fruit picking process, reduces economic losses caused by damage, and provides a scientific basis for farmers, helping them to formulate reasonable picking and processing strategies, thereby improving the market competitiveness of oil-tea fruit. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 It is a schematic diagram of the overall method flow of the present invention; Figure 2 It is a schematic diagram of the overall structure of the device of the present invention. DETAILED DESCRIPTION

[0017] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with specific embodiments.

[0018] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in the present invention should be understood by people with ordinary skills in the field to which the present invention belongs. The words "first", "second" and similar words used in the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. "Include" or "comprise" and similar words mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0019] Example: See also Figure 1 , the present invention provides a technical solution: A method for assessing fruit damage during tea-oil fruit picking, comprising the following specific steps: Step 1: Obtain several camellia fruit samples with known physiological characteristic parameters of camellia fruit, obtain spectral image data of the camellia fruit samples through a multispectral imaging device, preprocess the acquired spectral image data to obtain sample spectral image data, and map the sample spectral image data to the corresponding physiological characteristic parameters of camellia fruit one by one to generate a training sample data set, wherein the physiological characteristic parameters include water content, acetic acid content, ethanol content and butyric acid content.

[0020] Acquiring spectral image data of the Camellia oleifera fruit sample by a multispectral imaging device, wherein the wavelength range of the multispectral imaging device is 1400 nanometers to 2500 nanometers; Water has obvious absorption peaks in the near-infrared region (NIR), usually concentrated around 1450nm and 1940nm. The absorption peaks of water are mainly related to its molecular vibration mode, reflecting the water content in plant tissues. Fruits with high water content have lower reflectivity at these wavelengths.

[0021] The absorption peak of acetic acid usually appears around 1700nm, and may also involve the 1400nm band. The carboxyl group (-COOH) in the acetic acid molecule will produce absorption in a specific band, reflecting the presence and concentration of acetic acid.

[0022] The main absorption peaks of ethanol usually appear near 1450nm and 1800nm. The vibration of the -OH (hydroxyl) and -CH (alkyl) bonds in the ethanol molecule will cause its characteristic absorption in the near-infrared region, which can be used to quantitatively analyze the ethanol content.

[0023] The absorption characteristics of butyric acid are usually around 1700nm and 1400nm. The carboxyl group and carbon chain structure in the butyric acid molecule will cause absorption at specific wavelengths, which can indicate its content in the fruit.

[0024] After determining the characteristic band of each physiological characteristic parameter, the reflectance data of the characteristic band is extracted from the spectral data, and the physiological characteristic parameters are associated with the reflectance within the band to analyze the content of each physiological characteristic parameter.

[0025] The spectral image data of the camellia fruit samples are acquired by a multispectral imaging device, and the spectral image data of the camellia fruit samples are subjected to image preprocessing, wherein the preprocessing includes image enhancement and denoising preprocessing, wherein a wavelet transform denoising method is used to perform denoising on each spectral image of the camellia fruit sample, and a bilateral filter is used to perform image enhancement preprocessing on the spectral image of each camellia fruit sample.

[0026] The method for performing noise reduction and enhancement processing on the spectral image of the collected oil-tea camellia fruit sample is: using the wavelet transform denoising method to denoise the distortion-corrected image, and the specific steps of the wavelet transform denoising method include: decomposing the distortion-corrected image through wavelet transform to obtain the wavelet coefficients of the image at different scales and directions; performing threshold processing on the wavelet coefficients, setting the low-amplitude wavelet coefficients to zero, and retaining the high-amplitude wavelet coefficients; performing inverse transform on the wavelet coefficients after threshold processing, reconstructing the processed coefficients into an image, and completing the image denoising processing; Bilateral filtering was used to enhance the details of the spectral image of the Camellia oleifera fruit sample. The specific filter transformation was based on the formula: ; In the formula, is the coordinate vector in the image coordinate system, is the coordinate vector The gray value at Gray value The gray value after bilateral filtering transformation is are all Gaussian functions, where The formula is: ; ; In the formula, is the coordinate vector in the image coordinate system, is the coordinate vector The gray value at and They are The standard deviation of .

[0027] The method for generating the training sample data set is as follows: the sample spectral image data is mapped one by one with the physiological characteristic parameters of the corresponding tea fruit to form a corresponding grid, and the formed grid is recorded as the training sample data set.

[0028] Step 2: Based on the data in the training sample data set, a convolutional neural network model is established. The sample spectral image data in the training sample data set is used as the input of the convolutional neural network model, and the physiological characteristic parameters of tea fruit corresponding to the sample spectral image are used as labels to train the convolutional neural network model to obtain a physiological characteristic parameter prediction model.

[0029] Based on the data in the training sample data set, a convolutional neural network model is established, and based on the convolutional neural network, a physiological characteristic parameter prediction model is established, wherein the convolutional neural network consists of an input layer, a convolutional layer, a pooling layer, a fully connected layer and an output layer, wherein the activation function in the convolutional layer is function, The specific expression of the function is: ; in, Indicates The corresponding convolutional layers, It means the The corresponding convolutional layer The first training sample image eigenvalues, where is the index of the convolutional layer, is the index of the training sample image, is the index of the feature value in the training sample image; For the fully connected layer, the number of neurons in the fully connected layer is set to 32, the initial neural network learning rate is set to 0.001, and the number of training rounds is 200; The trained physiological characteristic parameter prediction model takes as input the spectral image data of the tea fruit, and outputs the predicted values ​​of the physiological characteristic parameters of the corresponding tea fruit.

[0030] Convolutional neural networks can automatically extract useful features from input spectral images without manually designing feature extraction algorithms. Traditional feature extraction methods often rely on expert experience and may miss some important information. CNN (convolutional neural network) can gradually build high-level abstract features from low-level features through multi-layer convolution and pooling operations, ensuring that the model can capture the relationship between complex physiological feature parameters.

[0031] Spectral image data usually has high-dimensional characteristics and contains rich spectral information. Traditional machine learning methods may face the "dimensionality disaster" problem when processing high-dimensional data, resulting in overfitting. CNN can effectively manage high-dimensional data and reduce the complexity of the model through its hierarchical structure and weight sharing mechanism, thereby improving the generalization ability of the model.

[0032] CNN has spatial invariance, that is, it has a certain robustness to transformations such as translation and scaling of the input image. This feature enables the model to better adapt to small changes in different spectral images, thereby improving the stability and accuracy of the prediction. For biological samples such as Camellia oleifera, changes in lighting, posture, and surface features may affect the performance of spectral images, and CNN can effectively reduce these interferences.

[0033] Step 3: During the picking process, the spectral image of the camellia oil fruit to be tested is obtained, and at the same time, the visible light true color image of the camellia oil fruit to be tested is collected; after preprocessing the spectral image of the camellia oil fruit to be tested, the image is input into the trained physiological characteristic parameter prediction model to obtain the predicted values ​​of the physiological characteristic parameters of the camellia oil fruit to be tested; and the appearance characteristic parameters of the camellia oil fruit to be tested are extracted according to the visible light true color image of the camellia oil fruit to be tested.

[0034] The specific method of extracting the appearance characteristic parameters of the oil-tea fruit to be detected from the visible light true color image of the oil-tea fruit to be detected includes: the contour perimeter of the oil-tea fruit refers to the total length of the boundary of the object in the image, which can usually be obtained using an edge detection algorithm, using methods such as Canny edge detection or Sobel operator to detect the edge, and then using a contour detection function. The edge detection and contour detection methods are also used to obtain the contour area of ​​the oil-tea fruit.

[0035] The color contrast of tea fruit can be obtained by the following steps: convert the image from RGB color space to HSV or Lab color space to better analyze the color, use the function in OpenCV to calculate the histogram, and then use it to calculate the color contrast. The color contrast of tea fruit specifically refers to the difference between the gray value of each pixel and the adjacent pixel in the identified tea fruit target image, and the average value of the difference of all pixels is taken as the contrast of tea fruit.

[0036] Step 4: Calculate and generate the influence coefficient of physiological damage of the camellia fruit based on the predicted values ​​of the physiological characteristic parameters of the camellia fruit to be tested; calculate and generate the influence coefficient of external damage of the camellia fruit based on the appearance characteristic parameters of the camellia fruit to be tested and the skin roughness of the camellia fruit to be tested, wherein the appearance characteristic parameters of the camellia fruit to be tested include the contour circumference, contour area and color contrast of the camellia fruit.

[0037] According to the predicted values ​​of the physiological characteristic parameters of the oil-tea camellia fruit to be tested, the influence coefficient of physiological damage to the oil-tea camellia fruit is calculated and generated. The formula for calculating the influence coefficient of physiological damage to the oil-tea camellia fruit is: ; In the formula, is the physiological damage influence coefficient of Camellia oleifera fruit, is the predicted value of water content of the oil-tea camellia fruit to be tested, is the predicted value of ethanol content in the oil-tea camellia fruit to be tested, is the predicted value of acetic acid content in the oil-tea camellia fruit to be tested, It is the predicted value of butyric acid content in the Camellia oleifera fruit to be tested.

[0038] It should be noted that the physiological damage coefficient of tea fruit Used to characterize the damage inside the oil-tea camellia fruit, among which the influence coefficient of physiological damage of the oil-tea camellia fruit The larger the value, the more serious the damage to the tea oil fruit.

[0039] Fresh fruits usually contain a high amount of water, which keeps the fruit crisp and tender with a good taste. Rotten fruits have less water and often appear shriveled, soft and rotten, losing their original texture and flavor. Fruit rot is usually caused by the activities of microorganisms such as bacteria and fungi. These microorganisms need water to survive and reproduce, resulting in a decrease in the water content of the fruit. Therefore, the predicted value of the water content of the oil-tea fruit to be tested is The influence coefficient of physiological damage on camellia oleifera fruit Inversely proportional, logarithmic function The introduction of The excessive influence of the fruit makes the damage assessment more balanced and better reflects the actual condition of the fruit.

[0040] Acetic acid is an organic acid produced during the fermentation process of fruits. It is usually related to cell membrane damage and abnormal metabolic processes of fruits. Increased acetic acid content usually indicates fruit decay or over-ripening. When fruits are damaged or infected by pathogens, the integrity of the cell wall will be affected, resulting in higher acetic acid concentrations. Therefore, acetic acid, as a marker of damage, can effectively reflect the physiological damage state of fruits. Therefore, the predicted value of acetic acid content in the oil-tea camellia fruit to be tested is The influence coefficient of physiological damage on camellia oleifera fruit Butyric acid is a strong volatile organic acid that is usually produced during fruit decay and deterioration. The presence of butyric acid is usually related to fruit corruption and anaerobic fermentation. When fruit cells are damaged or infected by microorganisms, the production of butyric acid increases, indicating that the fruit has been irreversibly damaged. Therefore, the increase in butyric acid content can directly indicate the degree of physiological damage to the fruit, so the exponential function The impact of these components is quickly magnified, especially when their content is high, emphasizing the potential risk of fruit damage. Acetic acid and butyric acid are volatile acids that may be produced during fruit ripening and damage, and are usually closely related to the health of the fruit. The square root of the sum of the squares of acetic acid and butyric acid content is calculated. , reflecting the combined influence of these two components.

[0041] Ethanol is one of the products of fruit fermentation and metabolism, usually produced in an oxygen-deficient environment. The increase of ethanol usually indicates that the fruit is in a fermentation state, which may be caused by cell damage or disease. High ethanol content not only reflects the health of the fruit, but also may affect the flavor and quality of the fruit. Therefore, the change of ethanol content can be used as an important indicator of physiological damage, so the predicted value of ethanol content of the oil-tea camellia fruit to be tested is The influence coefficient of physiological damage on camellia oleifera fruit Proportional, introduction of square term The higher ethanol content The contribution of 2.5% to 1.5% was more significant, reflecting changes in the health of the fruit.

[0042] Based on the appearance characteristic parameters of the oil-tea camellia fruit to be tested and the surface roughness of the oil-tea camellia fruit to be tested, the external damage influence coefficient of the oil-tea camellia fruit is calculated and generated. The formula for calculating the external damage influence coefficient of the oil-tea camellia fruit is: ; In the formula, is the influence coefficient of external damage to camellia oleifera fruit, is the color contrast of the tea fruit to be tested, is the roughness of the epidermis of the oil-tea camellia fruit to be tested, is the roundness of the outline of the tea fruit to be tested.

[0043] It should be noted that the external damage influence coefficient of Camellia oleifera fruit It is used to characterize the abnormality of the external shape and structure of Camellia oleifera fruit, among which the external damage influence coefficient of Camellia oleifera fruit is The larger the value, the more abnormal the external shape of the tea fruit is and the greater the damage it suffers.

[0044] In the case of physiological damage or pathological conditions, the fruit may show color changes such as fading, darkening or spots, which can directly reflect the health of the fruit. Therefore, changes in color contrast can be used as an indicator of external damage to the fruit. When fading, darkening or spots appear, the color contrast of the surface of the oil-tea camellia fruit will increase. Therefore, the color contrast of the oil-tea camellia fruit to be tested The influence coefficient of external damage on camellia oleifera fruit Proportional, using the natural logarithm function The effect of color contrast may be smoothed so that at lower contrasts the change in the damage influence coefficient is relatively small, while at higher contrasts the effect increases.

[0045] Skin roughness refers to the degree of undulation of the fruit surface texture, and is usually used to describe the touch and appearance of the fruit. The rougher the skin, the more damage and defects may exist on the outside of the fruit. For example, mechanical damage to the fruit during growth, infection by pests and diseases, or environmental factors can cause the skin to be uneven. Therefore, the skin roughness of the oil-tea fruit to be tested should be The influence coefficient of external damage on camellia oleifera fruit Proportional, through the square root function Can reduce the roughness This setting ensures that slight changes in roughness will not cause excessive fluctuations in the damage influence coefficient, while larger changes in roughness will significantly affect , which helps to accurately assess the external damage of the fruit.

[0046] Roundness refers to the regularity of the fruit shape, which is usually used to describe the geometric shape of the fruit. The ideal fruit shape should be close to a circle. Low roundness, that is, irregular shape, usually indicates that the fruit is deformed by collision or squeezing during picking. Therefore, the roundness of the outline of the oil-tea camellia fruit to be tested is The influence coefficient of external damage on camellia oleifera fruit Inversely proportional to the circularity, The introduction of the formula can more significantly amplify the contribution of irregular shapes to the external damage influence coefficient.

[0047] The circularity of the contour of the oil-tea camellia fruit to be detected is calculated by the appearance characteristic parameters of the oil-tea camellia fruit to be detected, and the specific formula for calculating the circularity of the contour of the oil-tea camellia fruit to be detected is: ; In the formula, is the outline area of ​​the tea fruit to be tested, is the circumference of the camellia oleifera fruit to be tested.

[0048] The roughness of the epidermis of the oil-tea camellia fruit to be tested is The specific acquisition method is: the epidermal roughness of the oil-tea fruit to be tested is tested by laser scanning. During the scanning process, the laser will form point cloud data on the surface of the fruit to record the shape and microscopic features of the surface. The specific steps include adjusting the working parameters of the laser scanner such as scanning speed, laser power and resolution according to the size and surface characteristics of the fruit, reconstructing the surface model of the oil-tea fruit through point cloud data, generating a three-dimensional surface image, and using the reconstructed three-dimensional surface model to calculate the epidermal roughness value using roughness analysis software.

[0049] Step 5: Based on the obtained physiological damage influence coefficient and external damage influence coefficient of the oil-tea fruit, a fruit damage assessment index is calculated comprehensively, and the generated fruit damage assessment index is compared with the damage degree judgment threshold. According to different comparison results, corresponding damage status judgment results are issued, among which the damage degree judgment threshold is dynamically corrected according to the environmental humidity of the place where the oil-tea fruit is picked.

[0050] According to the obtained physiological damage influence coefficient and external damage influence coefficient of Camellia oleifera fruit, the fruit damage assessment index is calculated comprehensively, and the formula for calculating the fruit damage assessment index is: ; In the formula, is the fruit damage assessment index, and are the weight coefficients of the influence coefficient of physiological damage of Camellia oleifera fruit and the influence coefficient of external damage of Camellia oleifera fruit, respectively. and and Both are greater than 0.

[0051] It should be noted that the fruit damage assessment index Used to comprehensively indicate the damage of tea fruit, including the fruit damage assessment index The larger the value, the more serious the fruit damage. As mentioned above, the influence coefficient of physiological damage to camellia fruit and the influence coefficient of external damage to camellia fruit are related to the fruit damage assessment index. The relative relationship between them will not be elaborated here.

[0052] where the exponential function The influence coefficient of physiological damage on the fruit damage assessment index of Camellia oleifera fruit The significant impact of The form of emphasizes the effect of external damage on the overall quality of the fruit. The square term makes the effect of external damage on the overall evaluation more significant.

[0053] Physiological damage is usually a key indicator of fruit quality, affecting its taste, nutritional content and shelf life. External damage can indirectly lead to physiological damage, so external damage is less important than physiological damage, but it still significantly affects the market value and appearance of the fruit. Therefore, it is necessary to set and and Both are greater than 0.

[0054] The generated fruit damage assessment index is compared with the damage degree judgment threshold, and the corresponding damage status judgment result is issued according to different comparison results. The specific judgment logic is as follows: when When the damage is detected, the tea fruit to be tested is judged to be first-grade damaged, indicating that the tea fruit to be tested cannot meet the needs of eating and processing; when When the damage is detected, the oil-tea camellia fruit to be tested is judged to be of the second-level damage, which means that the oil-tea camellia fruit to be tested cannot be sold directly and is used for secondary processing and sales; when When the oil-tea camellia fruit to be tested is judged to be level three damaged, it means that the oil-tea camellia fruit to be tested is intact and can be sold directly; in is the damage degree judgment threshold, which is obtained by dynamically correcting the ambient humidity at the location where the oil-tea fruit is picked. The calculation is based on the formula: ; In the formula, is the initial value of the damage degree judgment threshold, The environmental humidity of the tea fruit picking area. For reference humidity.

[0055] Ambient humidity will affect the prediction of fruit moisture content by spectrum. The higher the ambient humidity, the higher the predicted value of fruit moisture content by spectrum analysis, making the fruit damage assessment index The value will deviate from the actual situation and become smaller. Therefore, the humidity of the environment where the oil-tea fruit is picked is inversely proportional to the threshold for judging the degree of damage. The value is corrected accordingly.

[0056] The initial value of the damage degree judgment threshold Can be set through expert experience, reference humidity Generally set to to between.

[0057] See also Figure 2The present invention also provides a device for assessing damage to fruits during the picking process of oil-tea fruit. The device for assessing damage to fruits during the picking process of oil-tea fruit is used to execute the above-mentioned method for assessing damage to fruits during the picking process of oil-tea fruit, and comprises: A training data acquisition module is used to obtain several camellia fruit samples with known physiological characteristic parameters of camellia fruit, obtain spectral image data of the camellia fruit samples through a multispectral imaging device, pre-process the obtained spectral image data to obtain sample spectral image data, and map the sample spectral image data to the corresponding physiological characteristic parameters of the camellia fruit one by one to generate a training sample data set, wherein the physiological characteristic parameters include water content, acetic acid content, ethanol content and butyric acid content; The prediction model training module is used to establish a convolutional neural network model based on the data in the training sample data set, use the sample spectral image data in the training sample data set as the input of the convolutional neural network model, and use the physiological characteristic parameters of the oil-tea camellia fruit corresponding to the sample spectral image as labels to train the convolutional neural network model to obtain a physiological characteristic parameter prediction model; An appearance feature extraction module is used to obtain a spectral image of the oil-tea camellia fruit to be detected during the picking process, and simultaneously collect a visible light true color image of the oil-tea camellia fruit to be detected, and after pre-processing the spectral image of the oil-tea camellia fruit to be detected, input it into a trained physiological characteristic parameter prediction model to obtain a predicted value of the physiological characteristic parameter of the oil-tea camellia fruit to be detected, and extract the appearance feature parameters of the oil-tea camellia fruit to be detected based on the visible light true color image of the oil-tea camellia fruit to be detected; A damage factor analysis module is used to calculate and generate a camellia fruit physiological damage influence coefficient based on the predicted value of the camellia fruit physiological characteristic parameter to be detected, and to calculate and generate a camellia fruit external damage influence coefficient based on the appearance characteristic parameters of the camellia fruit to be detected and the epidermis roughness of the camellia fruit to be detected, wherein the appearance characteristic parameters of the camellia fruit to be detected include the camellia fruit contour circumference, contour area and color contrast; The comprehensive damage assessment module is used to comprehensively calculate the fruit damage assessment index based on the obtained physiological damage influence coefficient and external damage influence coefficient of the oil-tea fruit, compare the generated fruit damage assessment index with the damage degree judgment threshold, and issue corresponding damage status judgment results based on different comparison results, among which the damage degree judgment threshold is dynamically corrected by the environmental humidity of the place where the oil-tea fruit is picked.

[0058] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0059] The above embodiments may be implemented in whole or in part by software, hardware, firmware or any other combination thereof. When implemented by software, the above embodiments may be implemented in whole or in part in the form of a computer program product. Those skilled in the art may appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein may be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed by hardware or software methods depends on the specific application and design constraints of the technical solution.

[0060] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, and may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0061] The above description is only a specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application.

Claims

1. A method for assessing fruit damage during tea-oil fruit picking, characterized in that: The specific steps include: Acquire several camellia fruit samples with known physiological characteristic parameters of camellia fruit, acquire spectral image data of the camellia fruit samples by using a multispectral imaging device, pre-process the acquired spectral image data to obtain sample spectral image data, and map the sample spectral image data to the corresponding physiological characteristic parameters of camellia fruit one by one to generate a training sample data set, wherein the physiological characteristic parameters include water content, acetic acid content, ethanol content and butyric acid content; Based on the data in the training sample data set, a convolutional neural network model is established, the sample spectral image data in the training sample data set is used as the input of the convolutional neural network model, and the physiological characteristic parameters of tea fruit corresponding to the sample spectral image are used as labels to train the convolutional neural network model to obtain a physiological characteristic parameter prediction model; During the picking process, a spectral image of the oil-tea camellia fruit to be detected is obtained, and at the same time, a visible light true color image of the oil-tea camellia fruit to be detected is collected, and after preprocessing the spectral image of the oil-tea camellia fruit to be detected, the image is input into the trained physiological characteristic parameter prediction model to obtain the predicted value of the physiological characteristic parameter of the oil-tea camellia fruit to be detected, and the appearance characteristic parameters of the oil-tea camellia fruit to be detected are extracted according to the visible light true color image of the oil-tea camellia fruit to be detected; The physiological damage influence coefficient of the camellia fruit is calculated based on the predicted value of the physiological characteristic parameter of the camellia fruit to be tested, and the external damage influence coefficient of the camellia fruit is calculated based on the appearance characteristic parameter of the camellia fruit to be tested and the surface roughness of the camellia fruit to be tested, wherein the appearance characteristic parameter of the camellia fruit to be tested includes the contour circumference, contour area and color contrast of the camellia fruit; Based on the obtained physiological damage influence coefficient and external damage influence coefficient of tea oil fruit, the fruit damage assessment index is calculated comprehensively, and the generated fruit damage assessment index is compared with the damage degree judgment threshold. According to different comparison results, the corresponding damage status judgment result is issued, among which the damage degree judgment threshold is dynamically corrected according to the environmental humidity of the place where the tea oil fruits are picked.

2. The method for assessing fruit damage during the picking process of tea fruit according to claim 1, characterized in that: Acquiring spectral image data of the Camellia oleifera fruit sample by a multispectral imaging device, wherein the wavelength range of the multispectral imaging device is 1400 nanometers to 2500 nanometers; Acquiring spectral image data of the camellia fruit sample by a multispectral imaging device, and performing image preprocessing on the spectral image data of the camellia fruit sample, wherein the preprocessing includes image enhancement and denoising preprocessing, wherein a wavelet transform denoising method is used to perform denoising on each spectral image of the camellia fruit sample, and a bilateral filter is used to perform image enhancement preprocessing on each spectral image of the camellia fruit sample; The method for generating the training sample data set is as follows: the sample spectral image data is mapped one by one with the physiological characteristic parameters of the corresponding tea fruit to form a corresponding grid, and the formed grid is recorded as the training sample data set.

3. The method for assessing fruit damage during the picking process of tea oil fruits according to claim 2, characterized in that: Based on the data in the training sample data set, a convolutional neural network model is established, and based on the convolutional neural network, a physiological characteristic parameter prediction model is established, wherein the convolutional neural network consists of an input layer, a convolutional layer, a pooling layer, a fully connected layer and an output layer, wherein the activation function in the convolutional layer is function, The specific expression of the function is: ； in, Indicates The corresponding convolutional layers, It means the The corresponding convolutional layer The first training sample image eigenvalues, where is the index of the convolutional layer, is the index of the training sample image, is the index of the feature value in the training sample image; For the fully connected layer, the number of neurons in the fully connected layer is set to 32, the initial neural network learning rate is set to 0.001, and the number of training rounds is 200; The trained physiological characteristic parameter prediction model takes as input the spectral image data of the tea fruit, and outputs the predicted values ​​of the physiological characteristic parameters of the corresponding tea fruit.

4. The method for assessing fruit damage during the picking process of oil-tea camellia fruit according to claim 1, characterized in that: According to the predicted values ​​of the physiological characteristic parameters of the oil-tea camellia fruit to be tested, the influence coefficient of physiological damage to the oil-tea camellia fruit is calculated and generated. The formula for calculating the influence coefficient of physiological damage to the oil-tea camellia fruit is: ； In the formula, is the physiological damage influence coefficient of Camellia oleifera fruit, is the predicted value of water content of the oil-tea camellia fruit to be tested, is the predicted value of ethanol content in the oil-tea camellia fruit to be tested, is the predicted value of acetic acid content in the oil-tea camellia fruit to be tested, It is the predicted value of butyric acid content in the Camellia oleifera fruit to be tested.

5. The method for assessing fruit damage during the picking process of oil-tea camellia fruit according to claim 4, characterized in that: Based on the appearance characteristic parameters of the oil-tea camellia fruit to be tested and the surface roughness of the oil-tea camellia fruit to be tested, the external damage influence coefficient of the oil-tea camellia fruit is calculated and generated. The formula for calculating the external damage influence coefficient of the oil-tea camellia fruit is: ； In the formula, is the influence coefficient of external damage to camellia oleifera fruit, is the color contrast of the tea fruit to be tested, is the roughness of the skin of the Camellia oleifera fruit to be tested, is the roundness of the outline of the oil-tea camellia fruit to be tested; The circularity of the contour of the oil-tea camellia fruit to be detected is calculated by the appearance characteristic parameters of the oil-tea camellia fruit to be detected, and the specific formula for calculating the circularity of the contour of the oil-tea camellia fruit to be detected is: ； In the formula, is the outline area of ​​the tea fruit to be tested, is the circumference of the camellia oleifera fruit to be tested.

6. The method for assessing damage to tea fruit during picking according to claim 5, characterized in that: The roughness of the epidermis of the oil-tea camellia fruit to be tested is The specific acquisition method is: the skin roughness of the oil-tea fruit to be tested is detected by laser scanning. During the scanning process, the laser will form point cloud data on the surface of the fruit to record the shape and microscopic features of the surface.

7. The method for assessing damage to tea fruit during picking according to claim 6, characterized in that: According to the obtained physiological damage influence coefficient and external damage influence coefficient of Camellia oleifera fruit, the fruit damage assessment index is calculated comprehensively, and the formula for calculating the fruit damage assessment index is: ； In the formula, is the fruit damage assessment index, and are the weight coefficients of the influence coefficient of physiological damage of Camellia oleifera fruit and the influence coefficient of external damage of Camellia oleifera fruit, respectively. and and All are greater than 0; The generated fruit damage assessment index is compared with the damage degree judgment threshold, and the corresponding damage status judgment result is issued according to different comparison results. The specific judgment logic is as follows: when When the damage is detected, the tea fruit to be tested is judged to be first-grade damaged, indicating that the tea fruit to be tested cannot meet the needs of eating and processing; when When the damage is detected, the oil-tea camellia fruit to be tested is judged to be of the second-level damage, which means that the oil-tea camellia fruit to be tested cannot be sold directly and is used for secondary processing and sales; when When the oil-tea camellia fruit to be tested is judged to be level three damaged, it means that the oil-tea camellia fruit to be tested is intact and can be sold directly; in is the damage degree judgment threshold, which is obtained by dynamically correcting the ambient humidity at the location where the oil-tea fruit is picked. The calculation is based on the formula: ； In the formula, is the initial value of the damage degree judgment threshold, The environmental humidity of the tea fruit picking area. For reference humidity.

8. A device for assessing damage to oil-tea fruit during picking, characterized in that: The device for assessing damage to fruits during the picking process of oil-tea fruit is used to execute the method for assessing damage to fruits during the picking process of oil-tea fruit according to any one of claims 1 to 7, comprising: A training data acquisition module is used to obtain several camellia fruit samples with known physiological characteristic parameters of camellia fruit, obtain spectral image data of the camellia fruit samples through a multispectral imaging device, pre-process the obtained spectral image data to obtain sample spectral image data, and map the sample spectral image data to the corresponding physiological characteristic parameters of the camellia fruit one by one to generate a training sample data set, wherein the physiological characteristic parameters include water content, acetic acid content, ethanol content and butyric acid content; The prediction model training module is used to establish a convolutional neural network model based on the data in the training sample data set, use the sample spectral image data in the training sample data set as the input of the convolutional neural network model, and use the physiological characteristic parameters of the oil-tea camellia fruit corresponding to the sample spectral image as labels to train the convolutional neural network model to obtain a physiological characteristic parameter prediction model; An appearance feature extraction module is used to obtain a spectral image of the oil-tea camellia fruit to be detected during the picking process, and simultaneously collect a visible light true color image of the oil-tea camellia fruit to be detected, and after pre-processing the spectral image of the oil-tea camellia fruit to be detected, input it into a trained physiological characteristic parameter prediction model to obtain a predicted value of the physiological characteristic parameter of the oil-tea camellia fruit to be detected, and extract the appearance feature parameters of the oil-tea camellia fruit to be detected based on the visible light true color image of the oil-tea camellia fruit to be detected; A damage factor analysis module is used to calculate and generate a camellia fruit physiological damage influence coefficient based on the predicted value of the camellia fruit physiological characteristic parameter to be detected, and to calculate and generate a camellia fruit external damage influence coefficient based on the appearance characteristic parameters of the camellia fruit to be detected and the epidermis roughness of the camellia fruit to be detected, wherein the appearance characteristic parameters of the camellia fruit to be detected include the camellia fruit contour circumference, contour area and color contrast; The comprehensive damage assessment module is used to comprehensively calculate the fruit damage assessment index based on the obtained physiological damage influence coefficient and external damage influence coefficient of the oil-tea fruit, compare the generated fruit damage assessment index with the damage degree judgment threshold, and issue corresponding damage status judgment results based on different comparison results, among which the damage degree judgment threshold is dynamically corrected by the environmental humidity of the place where the oil-tea fruit is picked.

Images