Method and system for evaluating quality of polysaccharide-rich solid-state fermentation feed
By combining machine vision technology and random forest model, the accuracy and efficiency issues of quality assessment of solid fermented feed rich in polysaccharides were solved, and more efficient automated assessment was achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- INNER MONGOLIA AGRICULTURAL UNIVERSITY
- Filing Date
- 2023-08-08
- Publication Date
- 2026-04-28
AI Technical Summary
In existing technologies, the quality assessment of polysaccharide-rich solid fermented feed mainly relies on human experience, which has low accuracy and efficiency.
Machine vision technology is used to acquire images of polysaccharide-rich solid fermented feed through image acquisition equipment, extract color and texture features, and classify the feed using a pre-trained random forest model to determine whether the feed is qualified or unqualified.
It improves the accuracy and efficiency of evaluating solid fermented feed rich in polysaccharides, avoids the bias of human sensory experience judgment, and achieves more efficient quality evaluation.
Smart Images
Figure CN117030704B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of image processing, and in particular to a method and system for evaluating the quality of solid-state fermented feed rich in polysaccharides. Background Technology
[0002] Machine vision technology is an emerging technology that has developed in recent years. It mainly combines image acquisition equipment with computer analysis equipment, using computers and cameras to replace the human brain and eyes to measure, identify, process, classify, segment, analyze and make decisions on target areas. It extracts information from objective things for analysis, finds certain patterns, and replaces some manual work through algorithm analysis.
[0003] In recent years, some scholars have utilized machine vision technology combined with electronic noses, spectral monitoring, and olfactory visualization to monitor and assess the fermentation process from multiple dimensions. For example, they have used spectrophotometers to measure the color of theaflavins in fermented black tea infusions and determined the appropriate fermentation time based on these changes; and they have used machine vision systems to monitor changes in the appearance and color of tea leaves, combined with near-infrared spectroscopy to monitor changes in the main chemical components of the tea during fermentation. However, the quality assessment of solid-state fermented feeds rich in polysaccharides still relies on manual methods based on expert experience, resulting in low accuracy and efficiency. Summary of the Invention
[0004] The purpose of this invention is to provide a method and system for evaluating the quality of solid fermented feed rich in polysaccharides, which can improve the accuracy and efficiency of evaluating the quality of solid fermented feed rich in polysaccharides.
[0005] To achieve the above objectives, the present invention provides the following solution:
[0006] A method for quality assessment of solid-state fermented feed rich in polysaccharides, comprising:
[0007] Obtain solid-state fermented feed rich in polysaccharides;
[0008] Images of solid-state fermented feed rich in polysaccharides were acquired, and images of solid-state fermented feed rich in polysaccharides were obtained.
[0009] Extract the color and texture features of the polysaccharide-rich solid fermented feed image;
[0010] Based on the color and texture features, a pre-trained random forest model is used to determine the category of polysaccharide-rich solid fermented feed in the polysaccharide-rich solid fermented feed image; the category of polysaccharide-rich solid fermented feed is qualified or unqualified.
[0011] Optionally, images of polysaccharide-rich solid-state fermented feed are acquired to obtain images of polysaccharide-rich solid-state fermented feed, specifically including:
[0012] The polysaccharide-rich solid fermented feed is spread evenly in a petri dish; the petri dish is located on a stage inside an image acquisition box; the image acquisition box is equipped with an illumination source;
[0013] Images of the polysaccharide-rich solid fermented feed are obtained by acquiring images of the polysaccharide-rich solid fermented feed using an image acquisition device.
[0014] Optionally, the stage is a white matte stage; the inner wall of the image acquisition box is made of matte black light-absorbing material.
[0015] Optionally, the color features include RGB color parameters and HSV color parameters;
[0016] Extracting the color and texture features of the polysaccharide-rich solid-state fermented feed image, specifically including:
[0017] Identify the region of interest (ROI) in the image of the polysaccharide-rich solid-state fermented feed; the ROI includes both dense and porous regions of the material.
[0018] Based on the RGB color space, the RGB color features of the region of interest in the image of the polysaccharide-rich solid fermented feed are extracted; the RGB color features include red channel values, green channel values, and blue channel values.
[0019] Based on the RGB color characteristics of the region of interest, determine the RGB color parameters;
[0020] The RGB color features of the region of interest are converted to the HSV color space to obtain the HSV color features of the region of interest.
[0021] Determine the HSV color parameters based on the HSV color characteristics of the region of interest;
[0022] The gray-level co-occurrence matrix (GLCM) method is used to extract the GLCM of the region of interest.
[0023] Texture features are determined based on the gray-level co-occurrence matrix of the region of interest.
[0024] Optionally, the RGB color parameters include the mean of the red channel, the standard deviation of the red channel, the coefficient of variation of the red channel, the mean of the green channel, the standard deviation of the green channel, the coefficient of variation of the green channel, the mean of the blue channel, the standard deviation of the blue channel, and the coefficient of variation of the blue channel.
[0025] The HSV color parameters include the mean of the hue channel, the standard deviation of the hue channel, the coefficient of variation of the hue channel, the mean of the saturation channel, the standard deviation of the saturation channel, the coefficient of variation of the saturation channel, the mean of the lightness channel, the standard deviation of the lightness channel, and the coefficient of variation of the lightness channel.
[0026] Optionally, the texture features include: the mean of the second moment of the angle, the mean of the contrast, the mean of the correlation, the mean of the dissimilarity, the mean of the entropy, the mean of the homogeneity, the variance of the second moment of the angle, the variance of the contrast, the variance of the correlation, the variance of the dissimilarity, the variance of the entropy, and the variance of the homogeneity.
[0027] To achieve the above objectives, the present invention also provides the following solution:
[0028] A quality assessment system for solid-state fermented feed rich in polysaccharides includes:
[0029] The feed acquisition module is used to acquire solid-state fermented feed rich in polysaccharides;
[0030] The image acquisition module is used to acquire images of solid fermented feed rich in polysaccharides and obtain images of solid fermented feed rich in polysaccharides.
[0031] The feature extraction module, connected to the image acquisition module, is used to extract the color and texture features of the polysaccharide-rich solid fermented feed image;
[0032] The classification module, connected to the feature extraction module, is used to determine the category of polysaccharide-rich solid fermented feed in the image of polysaccharide-rich solid fermented feed based on the color features and texture features, using a pre-trained random forest model; the category of polysaccharide-rich solid fermented feed is qualified or unqualified.
[0033] Optionally, the image acquisition module includes: a plate, an image acquisition box, an illumination source, a stage, and an image acquisition device;
[0034] The stage and the illumination source are both located inside the image acquisition box;
[0035] The plate is located on the stage;
[0036] The polysaccharide-rich solid fermented feed was spread evenly in the petri dish;
[0037] The image acquisition device is used to acquire images of the polysaccharide-rich solid fermented feed to obtain images of the polysaccharide-rich solid fermented feed.
[0038] According to specific embodiments provided by the present invention, the present invention discloses the following technical effects:
[0039] This invention extracts color and texture features from images of polysaccharide-rich solid fermented feed and, based on these features, uses a pre-trained random forest model to assess the quality of polysaccharide-rich solid fermented feed in these images. By applying machine vision technology to the quality assessment of polysaccharide-rich solid fermented feed, this invention is more efficient than manual judgment and avoids biases caused by the lack of quantitative standards for human sensory experience, thus improving the accuracy of the assessment. Attached Figure Description
[0040] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0041] Figure 1 A flowchart of the method for evaluating the quality of polysaccharide-rich solid fermented feed provided by the present invention;
[0042] Figure 2 This is a schematic diagram of classification based on the random forest algorithm;
[0043] Figure 3 A schematic diagram illustrating the effects of random forest models with different classification schemes;
[0044] Figure 4 This is a schematic diagram of the polysaccharide-rich solid fermented feed quality evaluation system provided by the present invention.
[0045] Symbol explanation:
[0046] 1-Feed acquisition module, 2-Image acquisition module, 3-Feature extraction module, 4-Classification module. Detailed Implementation
[0047] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0048] The purpose of this invention is to provide a method and system for evaluating the quality of solid fermented feed rich in polysaccharides. By applying machine vision technology to the fermentation field, it is faster, provides more information, has more functions, and is more efficient than manual judgment, and avoids the deviation caused by the lack of quantitative standards for human sensory experience judgment.
[0049] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.
[0050] Example 1
[0051] like Figure 1 As shown in the figure, this embodiment provides a method for evaluating the quality of solid-state fermented feed rich in polysaccharides, including:
[0052] Step 100: Obtain solid fermented feed rich in polysaccharides.
[0053] Step 200: Acquire images of solid-state fermented feed rich in polysaccharides to obtain images of solid-state fermented feed rich in polysaccharides.
[0054] Solid fermented feed rich in polysaccharides is divided into powder feed and stem and leaf feed. Powder feed contains fermented main ingredients and fermented auxiliary ingredients, while stem and leaf feed contains fermented stems, fermented leaves, and fermented auxiliary ingredients.
[0055] Specifically, the polysaccharide-rich solid fermented feed is spread evenly in a petri dish. The spread should ideally completely cover the dish. The petri dish is located on a stage inside the image acquisition box. The petri dish is made of glass or plastic. The image acquisition box is equipped with a lighting source. The lighting source is a ring-shaped light source composed of LED lights (natural light, 100W). The stage is a matte white stage. The inner wall of the image acquisition box is made of matte black light-absorbing material.
[0056] Images of the polysaccharide-rich solid-state fermented feed are acquired using an image acquisition device, resulting in images of the polysaccharide-rich solid-state fermented feed. In one specific implementation, the image acquisition device is a smartphone.
[0057] Step 300: Extract the color and texture features of the polysaccharide-rich solid fermented feed image.
[0058] Specifically, the color features include RGB color parameters and HSV color parameters. Step 200 specifically includes:
[0059] (1) Determine the region of interest in the image of the polysaccharide-rich solid fermented feed. The region of interest includes the compacted material region and the porous material region. The extracted area of the compacted material region is 100 pixels × 100 pixels, and the extracted area of the porous material region is 30 pixels × 30 pixels.
[0060] (2) Using the RGB color space as a reference, extract the RGB color features of the region of interest in the image of the polysaccharide-rich solid fermented feed. The RGB color features include red channel values, green channel values, and blue channel values.
[0061] (3) Determine the RGB color parameters based on the RGB color characteristics of the region of interest.
[0062] Specifically, using the RGB color space of images of polysaccharide-rich solid-state fermented feed as a benchmark, the images were read using the `imread` function in Matlab software, and color features were extracted to obtain RGB color parameters. Then, smoothing and noise reduction were performed based on the mean-shift method. Finally, the RGB color features were converted into HSV color features.
[0063] In this embodiment, the RGB color parameters include the mean, standard deviation, and coefficient of variation of the red channel; the mean, standard deviation, and coefficient of variation of the green channel; the mean, standard deviation, and coefficient of variation of the blue channel; and the blue channel's mean, standard deviation, and coefficient of variation.
[0064] (4) Convert the RGB color features of the region of interest to the HSV color space to obtain the HSV color features of the region of interest. The HSV color features include hue, saturation and lightness.
[0065] Specifically, the rgb2hsv function is used to convert an RGB image to an HSV image. First, the red channel value R, green channel value G, and blue channel value B are all divided by 255 to convert them into R', G', and B' values between 0 and 1. Then, the red channel value R, green channel value G, and blue channel value B are converted into hue H, saturation S, and brightness V. The conversion formula is: V = max(R',G',B'); if V≠0, then S = [V-min(R',G',B')] / V; if V = 0, then S = 0; if V = R', then H = 60×(G'-B') / [V-min(R',G',B')]; if V = G', then H = 120+60×(B'-R') / [V-min(R',G',B')]; if V = B', then H = 240+60×(R'-G') / [V-min(R',G',B')]; if the calculated hue H is less than 0, add 360 to the value to get the final hue H.
[0066] (5) Determine the HSV color parameters based on the HSV color characteristics of the region of interest.
[0067] In this embodiment, the HSV color parameters include the mean of the hue channel, the standard deviation of the hue channel, the coefficient of variation of the hue channel, the mean of the saturation channel, the standard deviation of the saturation channel, the coefficient of variation of the saturation channel, the mean of the lightness channel, the standard deviation of the lightness channel, and the coefficient of variation of the lightness channel.
[0068] In summary, this invention extracts a total of 18 color features.
[0069] (6) The gray-level co-occurrence matrix method is used to extract the gray-level co-occurrence matrix of the region of interest.
[0070] (7) Determine the texture features based on the gray-level co-occurrence matrix of the region of interest.
[0071] By statistically analyzing the regions of interest (ROIs) of images of polysaccharide-rich solid-state fermented feed, two pixels with specific gray levels and a certain distance are identified to determine the gray-level co-occurrence matrix (GLCM). The "offset" of pixel pairs is taken from four different directions (0°, 45°, 90°, 135°). Six statistical measures based on the GLCM are then analyzed and calculated: second moment of angle, contrast, correlation, dissimilarity, entropy, and homogeneity. The mean and variance of these six statistical measures in the four directions constitute a 12-dimensional texture feature. The specific calculation formulas are as follows: Angular Second Moment (ASM) = sum(p(i,j)^2), Contrast Ratio (CON) = sum(p(i,j)*(ij).^2), Correlation Ratio (COR) = sum(p(i,j) / (1+(ij)^2)), Dissimilarity Ratio (DIS) = sum(p(i,j)*|ij|), Entropy Ratio (ENT) = sum(p(i,j)*(-log(p(i,j))), Homogeneity Ratio (HOM) = sum(p(i,j) / (1+(ij)^2)). Here, p(i,j) is the normalized gray-level co-occurrence matrix, representing the pixel pair of gray-level i and j, where i and j represent two different gray-level values.
[0072] In this embodiment, the texture features include: the mean of the second moment of the angle, the mean of the contrast, the mean of the correlation, the mean of the dissimilarity, the mean of the entropy, and the mean of the homogeneity; the variance of the second moment of the angle, the variance of the contrast, the variance of the correlation, the variance of the dissimilarity, the variance of the entropy, and the variance of the homogeneity.
[0073] Step 400: Based on the color features and texture features, a pre-trained random forest model is used to determine the category of polysaccharide-rich solid fermented feed in the polysaccharide-rich solid fermented feed image. The category of polysaccharide-rich solid fermented feed is either qualified or unqualified. Qualified indicates that the water-soluble polysaccharide content of the fermented feed is more than twice that of the unfermented feed, while unqualified indicates that the water-soluble polysaccharide content of the fermented feed is less than twice that of the unfermented feed.
[0074] During the training of the random forest model, the number of training sample images exceeded 2000. Five regions of interest were selected from each training sample image, representing both dense material regions and porous regions of interest. Step 200 was used to extract the color and texture features of these regions of interest. The RGB and HSV three-channel feature data were processed by mean, standard deviation, coefficient of variation, square, cube, natural logarithm, exponential, and reciprocal processing, yielding 24 feature data points for each. Texture features included raw value, mean, variance, homogeneity, contrast, dissimilarity, energy, correlation, and autocorrelation. Each texture feature underwent mean, standard deviation, and coefficient of variation processing, yielding 27 feature data points.
[0075] The construction of a random forest model based on color and texture feature parameters involves adding labels to the color and texture features, collecting all available data containing features and labels, and dividing it into training and test sets in an 8:2 ratio. A subset of samples is randomly selected from the training set to construct each decision tree, increasing model diversity through random sample selection. A subset is randomly selected from all features to train each decision tree. Different feature subsets are used to construct different classification methods, and the optimal random forest model for high-quality polysaccharide-rich solid-state fermented feed is selected using F1 score and accuracy. Figure 2 As shown.
[0076] Specifically, the validation process of the random forest model included: using a completely independent validation method, fermenting 30 samples again, and validating the model using sensory evaluation (conducted by at least three professional fermentation personnel) and the determination of the content of the main active ingredient, polysaccharide. Color and texture assessment and sensory evaluation were qualitative judgments, with results including both acceptable and unacceptable. The accuracy of the color and texture assessment was verified by comparing it with the sensory evaluation results; the accuracy was further verified by comparing it with the quantitative results of the main active ingredient, polysaccharide content. Finally, the high reliability and accuracy of the random forest model were demonstrated.
[0077] In the rapid evaluation of fermented feed quality based on machine vision technology, the present invention does not require complex pretreatment of samples, does not damage the sample components during analysis, only requires the acquisition of images of the sample components, without the addition of other reagents, and the analysis speed is fast.
[0078] To better understand the solution of this invention, wheat bran will be used as an example for explanation below.
[0079] (1) Construct an image dataset of wheat bran fermented at different times, with a total of 2750 valid images.
[0080] (2) Histograms were plotted on images of fermented wheat bran samples rich in polysaccharides, including histograms of the R, G, and B components in the RGB color space and histograms of the H, S, and V components in the HSV color space.
[0081] (3) Extract GLCM texture features from fermented wheat bran rich in polysaccharides.
[0082] (4) Random forest models were constructed using feature sets of RGB, HSV, GLCM, RGB+HSV, RGB+GLCM, HSV+GLCM, and RGB+HSV+GLCM. The classification scheme is shown in Table 1. The models were evaluated using F1 scores and accuracy from the sklearn database. The results of the F1 scores and accuracy are shown below. Figure 3 .from Figure 3 As can be seen, the evaluation model for the quality assessment of fermented wheat bran rich in polysaccharides is a mixed RGB+HSV+GLCM parameter random forest model.
[0083] Table 1 Classification schemes for random forest models
[0084] plan Sub-feature set Total number of features S1 RGB feature set 24 S2 HSV Feature Set 24 S3 GLCM feature set 27 S4 RGB+HSV feature set 48 S5 RGB+GLCM feature set 51 S6 HSV+GLCM feature set 51 S7 RGB+HSV+GLCM feature set 75
[0085] (5) The random forest model was completely independently validated.
[0086] Thirty images of polysaccharide-rich fermented wheat bran were independently and automatically evaluated using a random forest model. The results were compared with those from professional fermentation personnel and from chemometric determinations of the main active ingredient, polysaccharide content. The results are shown in Table 2. In Table 2, 1 indicates qualified, and 2 indicates unqualified. Unqualified samples have a polysaccharide content of less than 15%, while qualified samples have a polysaccharide content greater than 15%. The accuracy of the random forest model in evaluating the quality of polysaccharide-rich fermented wheat bran reached 86.67%, while the accuracy of manual judgment reached 73.33%. This demonstrates that the random forest model is more accurate than manual evaluation in assessing the quality of polysaccharide-rich fermented wheat bran, providing technical support for the standardized evaluation of polysaccharide-rich fermented wheat bran products.
[0087] Table 2. Quality evaluation results of polysaccharide-rich fermented wheat bran
[0088]
[0089]
[0090] Example 2
[0091] In order to implement the method corresponding to Embodiment 1 above and achieve the corresponding functions and technical effects, a quality evaluation system for solid fermented feed rich in polysaccharides is provided below.
[0092] like Figure 4 As shown, the polysaccharide-rich solid fermented feed quality assessment system provided in this embodiment includes: feed acquisition module 1, image acquisition module 2, feature extraction module 3, and classification module 4.
[0093] Among them, the feed acquisition module 1 is used to acquire solid fermented feed rich in polysaccharides.
[0094] Image acquisition module 2 is used to acquire images of solid fermented feed rich in polysaccharides, and obtain images of solid fermented feed rich in polysaccharides.
[0095] Specifically, the image acquisition module 2 includes: a plate, an image acquisition box, an illumination source, a stage, and an image acquisition device.
[0096] The stage and the illumination source are both located inside the image acquisition box.
[0097] The plate is located on the stage.
[0098] The polysaccharide-rich solid fermented feed was spread evenly in the petri dish.
[0099] The image acquisition device is used to acquire images of the polysaccharide-rich solid fermented feed to obtain images of the polysaccharide-rich solid fermented feed.
[0100] The feature extraction module 3 is connected to the image acquisition module 2. The feature extraction module 3 is used to extract the color features and texture features of the polysaccharide-rich solid fermented feed image.
[0101] The classification module 4 is connected to the feature extraction module 3. The classification module 4 is used to determine the category of polysaccharide-rich solid fermented feed in the image of polysaccharide-rich solid fermented feed based on the color features and the texture features, using a pre-trained random forest model. The category of polysaccharide-rich solid fermented feed is either qualified or unqualified.
[0102] Compared with the prior art, the quality evaluation system for polysaccharide-rich solid fermented feed provided in this embodiment has the same beneficial effects as the quality evaluation method for polysaccharide-rich solid fermented feed provided in Embodiment 1, and will not be repeated here.
[0103] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The same or similar parts between the various embodiments can be referred to each other.
[0104] This document uses specific examples to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. Furthermore, those skilled in the art will recognize that, based on the ideas of the present invention, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of the present invention.
Claims
1. A method for evaluating the quality of solid-state fermented feed rich in polysaccharides, characterized in that, The method for evaluating the quality of polysaccharide-rich solid-state fermented feed includes: Obtain solid-state fermented feed rich in polysaccharides; Images of solid-state fermented feed rich in polysaccharides were acquired, and images of solid-state fermented feed rich in polysaccharides were obtained. Extract the color and texture features of the polysaccharide-rich solid fermented feed image; Based on the color and texture features, a pre-trained random forest model is used to determine the category of polysaccharide-rich solid fermented feed in the polysaccharide-rich solid fermented feed image; the category of polysaccharide-rich solid fermented feed is qualified or unqualified.
2. The method for quality evaluation of polysaccharide-rich solid-state fermented feed according to claim 1, characterized in that, Images of polysaccharide-rich solid-state fermented feed were acquired, specifically including: The polysaccharide-rich solid fermented feed is spread evenly in a petri dish; the petri dish is located on a stage inside an image acquisition box; the image acquisition box is equipped with an illumination source; Images of the polysaccharide-rich solid fermented feed are obtained by acquiring images of the polysaccharide-rich solid fermented feed using an image acquisition device.
3. The method for quality evaluation of polysaccharide-rich solid-state fermented feed according to claim 2, characterized in that, The stage is a white matte stage; the inner wall of the image acquisition box is made of matte black light-absorbing material.
4. The method for quality evaluation of polysaccharide-rich solid-state fermented feed according to claim 1, characterized in that, The color features include RGB color parameters and HSV color parameters; Extracting the color and texture features of the polysaccharide-rich solid-state fermented feed image, specifically including: Identify the region of interest (ROI) in the image of the polysaccharide-rich solid-state fermented feed; the ROI includes both dense and porous regions of the material. Based on the RGB color space, the RGB color features of the region of interest in the image of the polysaccharide-rich solid fermented feed are extracted; the RGB color features include red channel values, green channel values, and blue channel values. Based on the RGB color characteristics of the region of interest, determine the RGB color parameters; The RGB color features of the region of interest are converted to the HSV color space to obtain the HSV color features of the region of interest. Determine the HSV color parameters based on the HSV color characteristics of the region of interest; The gray-level co-occurrence matrix (GLCM) method is used to extract the GLCM of the region of interest. Texture features are determined based on the gray-level co-occurrence matrix of the region of interest.
5. The method for quality evaluation of polysaccharide-rich solid-state fermented feed according to claim 4, characterized in that, The RGB color parameters include the mean, standard deviation, and coefficient of variation of the red channel; the mean, standard deviation, and coefficient of variation of the green channel; the mean, standard deviation, and coefficient of variation of the blue channel; and the blue channel. The HSV color parameters include the mean of the hue channel, the standard deviation of the hue channel, the coefficient of variation of the hue channel, the mean of the saturation channel, the standard deviation of the saturation channel, the coefficient of variation of the saturation channel, the mean of the lightness channel, the standard deviation of the lightness channel, and the coefficient of variation of the lightness channel.
6. The method for quality evaluation of polysaccharide-rich solid-state fermented feed according to claim 4, characterized in that, The texture features include: the mean of the second moment of the angle, the mean of the contrast, the mean of the correlation, the mean of the dissimilarity, the mean of the entropy, the mean of the homogeneity, the variance of the second moment of the angle, the variance of the contrast, the variance of the correlation, the variance of the dissimilarity, the variance of the entropy, and the variance of the homogeneity.
7. A quality evaluation system for solid-state fermented feed rich in polysaccharides, characterized in that, The polysaccharide-rich solid-state fermented feed quality assessment system includes: The feed acquisition module is used to acquire solid-state fermented feed rich in polysaccharides; The image acquisition module is used to acquire images of solid fermented feed rich in polysaccharides and obtain images of solid fermented feed rich in polysaccharides. The feature extraction module, connected to the image acquisition module, is used to extract the color and texture features of the polysaccharide-rich solid fermented feed image; The classification module, connected to the feature extraction module, is used to determine the category of polysaccharide-rich solid fermented feed in the image of polysaccharide-rich solid fermented feed based on the color features and texture features, using a pre-trained random forest model; the category of polysaccharide-rich solid fermented feed is qualified or unqualified.
8. The quality evaluation system for polysaccharide-rich solid-state fermented feed according to claim 7, characterized in that, The image acquisition module includes: a plate, an image acquisition box, an illumination source, a stage, and an image acquisition device; The stage and the illumination source are both located inside the image acquisition box; The plate is located on the stage; The polysaccharide-rich solid fermented feed was spread evenly in the petri dish; The image acquisition device is used to acquire images of the polysaccharide-rich solid fermented feed to obtain images of the polysaccharide-rich solid fermented feed.