Straw feed nutrition analysis system and method based on image recognition

Abstract

The invention discloses a straw feed nutrition analysis system and method based on image recognition, which specifically relates to the technical field of image analysis, and is used to solve the problem that the existing system cannot accurately capture the microscopic differences in straw fiber arrangement, resulting in insufficient accuracy of nutrition analysis results. The system collects image data of the straw surface through multi-directional optical imaging, and divides the image data into multiple analysis sub-areas. The system extracts fiber arrangement information based on morphological analysis to enhance texture features, and extracts fiber micro-difference indicators based on pixel distribution and neighborhood relationships. The system uses a multi-fractal model to perform complexity analysis on the distribution characteristics of the fiber micro-difference indicators to determine the randomness of the fiber arrangement pattern. The system dynamically analyzes the change trend of the fiber micro-difference indicators at different time sequences through a nonlinear method of phase space reconstruction to determine the offset amplitude of the fiber arrangement. The system comprehensively determines the stability of the fiber micro-difference indicators, and compares them with fiber component reference data to finally generate straw nutrition analysis data.

Description

Technical Field

[0001] The present invention relates to the technical field of image analysis, and more specifically, to a straw feed nutrition analysis system and method based on image recognition. Background Art

[0002] In the production and application of straw feed, accurate assessment of its fiber nutritional content is an important basis for determining its feeding value. At present, most technologies conduct nutritional assessment by directly detecting the main components of straw fiber (such as cellulose, hemicellulose, lignin, etc.). However, these methods usually rely on chemical analysis or simple surface texture image processing, ignoring the subtle differences in fiber arrangement at the microscopic level. Straws from different sources or with different treatment methods may have complex changes in their fiber arrangement patterns and distribution patterns. These subtle differences are often difficult to capture effectively using existing methods, but have a significant impact on the accuracy of nutritional assessment, especially when there are large differences between batches.

[0003] In the existing technology, there is a lack of efficient identification and evaluation methods for subtle differences in straw fiber arrangement, which leads to insufficient accuracy and stability of nutritional component analysis results. Summary of the invention

[0004] In order to overcome the above-mentioned defects of the prior art, the embodiments of the present invention provide a straw feed nutrition analysis system and method based on image recognition to solve the problems raised in the above-mentioned background technology.

[0005] To achieve the above object, the present invention provides the following technical solutions:

[0006] The straw feed nutrition analysis method based on image recognition comprises the following steps:

[0007] Perform multi-directional optical imaging on the straw surface to obtain straw image data, and divide the straw image data into multiple analysis sub-areas;

[0008] The fiber arrangement information of the sub-region is extracted based on morphological analysis to enhance the fiber texture characteristics, and the fiber micro-difference index is extracted based on pixel distribution and neighborhood relationship;

[0009] The complexity analysis of the distribution characteristics of fiber micro-difference indexes is carried out through image decomposition based on multi-fractal model to determine whether the randomness of fiber arrangement pattern in local area is normal.

[0010] The variation trend of fiber micro-difference index at different time sequences is analyzed by nonlinear method based on phase space reconstruction to judge whether the deviation amplitude of fiber arrangement is normal.

[0011] Based on the randomness of the fiber arrangement pattern in the local area and the judgment result of whether the deviation amplitude of the fiber arrangement is normal, it is judged whether the stability of the fiber micro-difference index meets the standard;

[0012] When the stability of the fiber micro-difference index meets the standard, the fiber micro-difference index is compared with the fiber composition reference data. After the fiber micro-difference index meets the preset standard, the straw nutritional analysis data is determined.

[0013] In a preferred embodiment, multi-directional optical imaging is performed on the straw surface to obtain straw image data, and the straw image data is divided into multiple analysis sub-areas, specifically including:

[0014] A plurality of optical imaging devices are arranged on the surface of the straw and a plurality of light sources are configured, and the image data of the straw in multiple directions are recorded by adjusting the focal length and shooting parameters of the optical imaging devices;

[0015] The multi-directional straw image data are stored and preprocessed to form a preprocessed image set, and the preprocessed image set is decomposed into multiple analysis sub-areas according to a preset division criterion.

[0016] In a preferred embodiment, the fiber arrangement information of the sub-region is extracted based on morphological analysis to enhance the fiber texture characteristics, and the fiber micro-difference index is extracted based on pixel distribution and neighborhood relationship, specifically including:

[0017] Morphological analysis is applied to the straw image data in the analysis sub-area, and fiber texture features are enhanced by selecting multi-scale structural elements, which are used to adapt to fiber textures of different thicknesses and directions.

[0018] The fiber arrangement information in the analysis sub-area is quantitatively analyzed based on the pixel distribution parameters, which are obtained by calculating the gray value variation characteristics of the local area;

[0019] The neighborhood relationship of fiber arrangement information is analyzed, and the neighborhood characteristic parameters of fiber arrangement are extracted based on the spatial correlation between local pixels and their surrounding pixels.

[0020] The fiber micro-difference index is extracted by combining pixel distribution parameters with neighborhood characteristic parameters. The fiber micro-difference index is used to characterize the microscopic difference characteristics of fiber arrangement.

[0021] In a preferred embodiment, the distribution characteristics of the fiber difference index are analyzed by complexity analysis based on the image decomposition of the multi-fractal model to determine whether the randomness of the fiber arrangement pattern in the local area is normal, which specifically includes:

[0022] Generate a two-dimensional distribution matrix by distributing the fiber differential indexes calculated in the analysis sub-area according to the spatial coordinates;

[0023] Decomposing the two-dimensional distribution matrix into multiple scale components based on the multi-fractal model;

[0024] The fractal dimension corresponding to each level of component is calculated to quantify the complexity of fiber arrangement at different scales in the two-dimensional distribution matrix;

[0025] According to the distribution law of fractal dimension at different scales, the complexity characteristic parameters representing the randomness of fiber arrangement are extracted. The complexity characteristic parameters include fractal spectrum width and fractal intensity.

[0026] The complexity characteristic parameters are compared with the preset standards to determine whether the randomness of the fiber arrangement pattern in the local area is normal.

[0027] In a preferred embodiment, the variation trend of the fiber differential index at different time sequences is analyzed by a nonlinear method based on phase space reconstruction to determine whether the deviation amplitude of the fiber arrangement is normal, specifically including:

[0028] The fiber micro-difference indexes at each time point are arranged in order to form a fiber micro-difference index time series;

[0029] The fiber differential index time series is mapped into a multi-dimensional state vector set by phase space reconstruction;

[0030] Applying nonlinear analysis methods to the multidimensional state vector set, the offset-related characteristic parameters are extracted. The offset-related characteristic parameters are used to quantify the variation of fiber arrangement in the time dimension.

[0031] The deviation-related characteristic parameters are compared with the preset reference range to determine whether the deviation amplitude of the fiber arrangement is normal.

[0032] In a preferred embodiment, a nonlinear analysis method is applied to the multidimensional state vector set to extract the offset-related characteristic parameters, specifically:

[0033] Calculate the Euclidean distance of the corresponding state vectors at adjacent time points as the local offset distance: ;in, Indicates time point The local offset distance, Indicates time point The state vector of Quantity, is the index of the component in the state vector, is the embedding dimension.

[0034] In a preferred embodiment, based on the randomness of the fiber arrangement pattern in the local area and the judgment result of whether the deviation amplitude of the fiber arrangement is normal, it is judged whether the stability of the fiber micro-difference index meets the standard, specifically including:

[0035] When the randomness of the fiber arrangement pattern in the local area is normal and the deviation amplitude of the fiber arrangement is normal, it is determined that the stability of the fiber micro-difference index meets the standard; otherwise, it is determined that the stability of the fiber micro-difference index does not meet the standard.

[0036] In a preferred embodiment, when the stability of the fiber difference index meets the standard, the fiber difference index is compared with the fiber component reference data, and after the fiber difference index meets the preset standard, the straw nutrition analysis data is determined, specifically including:

[0037] Selecting corresponding reference data from a preset fiber component reference database according to the attributes of the analysis sub-region;

[0038] Compare the fiber difference index with the selected reference data item by item to determine whether it meets the preset standards;

[0039] When all comparison items between the fiber differential index and the reference data meet the preset standards, the straw nutritional analysis data is generated.

[0040] On the other hand, the present invention provides a straw feed nutrition analysis system based on image recognition, including an optical imaging acquisition module, a fiber micro-difference extraction module, a random complex analysis module, an offset amplitude analysis module, a stable comprehensive judgment module and a nutrition data generation module;

[0041] Optical imaging acquisition module: performs multi-directional optical imaging acquisition on the straw surface to obtain straw image data, and divides the straw image data into multiple analysis sub-areas;

[0042] Fiber micro-difference extraction module: extracts fiber arrangement information of the analysis sub-region based on morphological analysis to enhance fiber texture features, and extracts fiber micro-difference indicators based on pixel distribution and neighborhood relationship;

[0043] Random Complex Analysis Module: Through image decomposition based on multi-fractal model, the distribution characteristics of fiber micro-difference index are analyzed for complexity, and the randomness of fiber arrangement pattern in the local area is judged to be normal;

[0044] Offset amplitude analysis module: Analyze the variation trend of fiber differential index at different time sequences through nonlinear method based on phase space reconstruction to determine whether the offset amplitude of fiber arrangement is normal;

[0045] Stability comprehensive judgment module: Based on the randomness of the fiber arrangement pattern in the local area and whether the deviation amplitude of the fiber arrangement is normal, it is judged whether the stability of the fiber micro-difference index meets the standard;

[0046] Nutritional data generation module: When the stability of the fiber differential index meets the standard, the fiber differential index is compared with the fiber composition reference data. After the fiber differential index meets the preset standard, the straw nutritional analysis data is determined.

[0047] The technical effects and advantages of the straw feed nutrition analysis system and method based on image recognition of the present invention are as follows:

[0048] 1. Through multi-directional optical imaging acquisition technology, the multi-dimensional texture information of the straw surface can be fully recorded, providing high-quality data input for subsequent analysis; morphological analysis combined with multi-scale structural elements can effectively enhance the fiber texture characteristics, making the microscopic differences more prominent. Further, based on the complexity analysis of the multi-fractal model and the dynamic analysis of phase space reconstruction, the dynamic characteristics of the fiber arrangement can be accurately evaluated from the two dimensions of randomness and offset amplitude of the fiber arrangement, ensuring that the extracted fiber micro-difference indicators can fully reflect the microscopic characteristics of the straw fiber.

[0049] 2. By comparing the fiber differential index with the fiber composition reference data item by item, it is possible to accurately determine the degree of match between the fiber arrangement characteristics and the expected standards, and achieve high-precision evaluation of the nutritional components of straw, especially for straw samples from different sources or different processing batches. The dual mechanism of stability judgment and data comparison significantly improves the accuracy and consistency of nutritional evaluation. Compared with traditional chemical analysis methods, it can not only achieve non-destructive testing, but also quickly process large-scale samples, significantly improving analysis efficiency, and at the same time providing a scientific basis for feed formula optimization and production process control. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 This is a schematic diagram of a straw feed nutrition analysis method based on image recognition according to the present invention;

[0051] Figure 2 It is a structural schematic diagram of the straw feed nutrition analysis system based on image recognition of the present invention. DETAILED DESCRIPTION

[0052] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0053] Embodiment 1: Figure 1 The present invention provides a straw feed nutrition analysis method based on image recognition, which comprises the following steps:

[0054] Multi-directional optical imaging is performed on the straw surface to obtain straw image data, and the straw image data is divided into multiple analysis sub-areas.

[0055] The fiber arrangement information of the analysis sub-region is extracted based on morphological analysis to enhance the fiber texture characteristics, and the fiber micro-difference index is extracted based on pixel distribution and neighborhood relationship.

[0056] The complexity analysis of the distribution characteristics of fiber micro-difference indexes is carried out through image decomposition based on multi-fractal model to determine whether the randomness of the fiber arrangement pattern in the local area is normal.

[0057] The variation trend of fiber differential index at different time sequences is analyzed by nonlinear method based on phase space reconstruction to judge whether the deviation amplitude of fiber arrangement is normal.

[0058] Based on the randomness of the fiber arrangement pattern in the local area and whether the deviation amplitude of the fiber arrangement is normal, it is judged whether the stability of the fiber micro-difference index meets the standard.

[0059] When the stability of the fiber micro-difference index meets the standard, the fiber micro-difference index is compared with the fiber composition reference data. After the fiber micro-difference index meets the preset standard, the straw nutritional analysis data is determined.

[0060] Multi-directional optical imaging is performed on the straw surface to obtain straw image data, and the straw image data is divided into multiple analysis sub-areas, including:

[0061] Multiple optical imaging devices are set up on the surface of the straw and multiple light sources are configured. By adjusting the focal length and shooting parameters of the optical imaging devices, multi-directional straw image data are recorded:

[0062] Among them, the configuration of the optical imaging device should ensure that different viewing angle data of the straw surface can be obtained from multiple directions to fully capture its surface texture characteristics; by configuring multiple light sources, it can ensure that light from different directions can illuminate the straw surface, avoid shadow effects, and improve image clarity and contrast.

[0063] The focal length of the optical imaging device is adjusted according to the surface characteristics and texture of the straw to ensure the acquisition of high-resolution image data, especially for fine fiber structures and surface irregularities; by setting shooting parameters such as exposure time, shutter speed and gain, image data can be stably collected under different environmental conditions to ensure the stability of image quality.

[0064] The multi-directional straw image data are stored and preprocessed to form a preprocessed image set, and the preprocessed image set is decomposed into multiple analysis sub-areas according to the preset division criteria:

[0065] The straw image data taken from multiple directions are stored and organized into a pre-processed image set, which contains straw images from different angles to ensure the comprehensiveness of the data.

[0066] Image denoising is performed to remove noise generated during the shooting process, enhance image clarity, and correct the image to adjust color and brightness to make it suitable for subsequent processing.

[0067] According to the preset division criteria (such as continuity of fiber structure, uniformity of image content, etc.), the processed image set is decomposed into multiple independent analysis sub-regions. These regions should be independent and representative, and be able to reflect the different fiber structure units of the straw.

[0068] Among them, the preset division criteria are standards formulated based on factors such as the fiber arrangement characteristics of the straw surface, image texture characteristics, and regional uniformity, and are used to divide the preprocessed image data into multiple analysis sub-regions. Specific criteria include: based on the continuity of fiber texture as the main basis, when the texture characteristics maintain a high correlation or consistency in a certain area, they are divided into the same sub-region; based on the color gradient change as an auxiliary basis, when the brightness or color distribution of the local area changes within a certain range, it is divided into independent sub-regions. For example, for a straw surface image, if the arrangement direction of a part of the fibers is basically consistent and the color gradient is relatively smooth, it can be used as an analysis sub-region; in another part of the image, if the fibers show obvious direction changes or color transitions, they should be divided into new analysis sub-regions; this division method ensures that each sub-region has consistency in features, which facilitates the subsequent extraction and analysis of fiber arrangement information.

[0069] Among them, the straw surface refers to the surface of the straw feed, and its acquisition method includes but is not limited to sampling from the straw feed. Straw feed samples are selected randomly or in batches to ensure representativeness and diversity. After sampling, the relatively complete part of the surface can be directly selected for analysis, or the sample surface can be cleaned to eliminate the influence of external contaminants and provide reliable surface conditions for subsequent optical imaging acquisition.

[0070] The fiber arrangement information of the sub-region is extracted based on morphological analysis to enhance the fiber texture characteristics, and the fiber micro-difference index is extracted based on pixel distribution and neighborhood relationship, including:

[0071] Morphological analysis is applied to the straw image data in the analysis sub-area, and the fiber texture features are enhanced by selecting multi-scale structural elements. The multi-scale structural elements are used to adapt to fiber textures of different thicknesses and directions:

[0072] For each analysis sub-region in the straw image data, morphological analysis is applied to enhance the fiber texture characteristics; the core of morphological analysis is to use multi-scale structural elements to process the fiber texture, where the multi-scale structural elements are geometric primitives composed of specific size and shape parameters, which are used to adapt to the fiber textures of different thicknesses and directions in the analysis sub-region.

[0073] Assume that the analysis sub-region in the straw image data is represented as ,in and are the horizontal and vertical coordinates of the image, Apply morphological operations, the formula is: ;in, Represents the image data after morphological enhancement; It is a multi-scale structural element; Indicates the scale of the structural element, and its value range is ; Indicates the direction parameter, the value range is ; represents the number of scales of the multi-scale structural element, The number of directions representing the multi-scale structural elements; Represents morphological opening or closing operations, used for denoising and enhancing fiber texture;

[0074] Based on By applying morphological operations, the fiber texture features at different scales and directions can be enhanced to ensure that the fiber arrangement information in the analyzed sub-region is fully preserved.

[0075] The fiber arrangement information in the analysis sub-area is quantitatively analyzed based on the pixel distribution parameters, which are obtained by calculating the gray value change characteristics of the local area:

[0076] For the image data after morphological enhancement, pixel distribution parameters are extracted based on the pixel grayscale distribution characteristics to quantitatively analyze the local changes in fiber arrangement information. The pixel distribution parameters are calculated based on the changes in grayscale values ​​in the local area to reflect the density and uniformity of the fiber texture.

[0077] For example, a local window is selected in the analysis subregion, and its center pixel coordinates are , the window size is , then the pixel distribution parameter is defined as: ;in, Represents the pixel distribution parameters of the window center point; Indicates the gray value of a pixel in the window; Represents the gray value of the center point in the window; Indicates the side length of the window; Indicates the horizontal offset of the current pixel in the local window relative to the upper left corner of the window, which is used to traverse each pixel of the window; Indicates the vertical offset of the current pixel in the local window relative to the upper left corner of the window, which is used to traverse each pixel of the window; Indicates the vertical coordinate of the center pixel of the local window, which is used to locate the specific position of the window in the image; Indicates the horizontal coordinate of the center pixel of the local window, which is used to locate the specific position of the window in the image.

[0078] The size of the pixel distribution parameter reflects the intensity of the grayscale change of the fiber arrangement, including: a larger parameter indicates that the grayscale value of the fiber arrangement in the local area changes more dramatically, which may indicate uneven fiber distribution or prominent texture differences; a smaller parameter indicates that the grayscale value changes more slowly in the local area, which may indicate a more uniform fiber distribution or weaker texture characteristics.

[0079] By calculating the pixel distribution parameters of all window center points in each analysis sub-area, the grayscale change characteristics of the fiber arrangement are quantified. By calculating the pixel distribution parameters of each window center point, the pixel distribution parameters corresponding to all window center points in the analysis sub-area can be aggregated to form a pixel distribution feature map in the area. This feature map reflects the grayscale change trend of the fiber arrangement in the local area, including the distribution of high grayscale change areas and low grayscale change areas, thereby quantifying the density and uniformity characteristics of the fiber arrangement.

[0080] The neighborhood relationship of the fiber arrangement information is analyzed, and the neighborhood characteristic parameters of the fiber arrangement are extracted based on the spatial correlation between the local pixel and its surrounding pixels:

[0081] Based on the spatial correlation characteristics between pixels, the neighborhood characteristic parameters of fiber arrangement are extracted, and the neighborhood relationship is used to reflect the structural consistency and spatial distribution characteristics of pixel points in the local area.

[0082] The neighborhood characteristic parameters of the pixel points are defined as: ;in, Represents the neighborhood feature parameters of the pixel point, Represents the total number of pixels in the neighborhood. Indicates the sequence number of the current pixel in the neighborhood. and Respectively represent Pixels relative to the center point The horizontal and vertical offsets.

[0083] The size of the neighborhood characteristic parameter reflects the local spatial correlation of the fiber arrangement, including: a larger parameter indicates that the correlation between the pixel point and its neighboring pixels is stronger, indicating that the fiber arrangement has high consistency and coherence; a smaller parameter indicates that the correlation between the pixel point and its neighboring pixels is weaker, indicating that the fiber arrangement may be more discrete or random.

[0084] The fiber micro-difference index is extracted by combining the pixel distribution parameters and the neighborhood characteristic parameters. The fiber micro-difference index is used to characterize the microscopic difference characteristics of fiber arrangement:

[0085] After the extraction of pixel distribution parameters and neighborhood feature parameters, the two are combined to generate a fiber micro-difference index, which is used to characterize the microscopic difference characteristics of fiber arrangement.

[0086] The fiber differential index is defined as: ;in, Indicates fiber micro-difference index, is the average value of the pixel distribution parameter, is the average value of the neighborhood characteristic parameters, and Represent the weights of pixel distribution parameters and neighborhood feature parameters respectively, and Both are greater than 0.

[0087] The fiber micro-difference index combines the pixel distribution parameters and the neighborhood characteristic parameters, and is used to characterize the microscopic difference characteristics of fiber arrangement. A larger fiber micro-difference index indicates that the fiber arrangement in the analysis sub-region is significantly different, which may be manifested as increased randomness of the fiber arrangement or lower local correlation, usually reflecting uneven fiber distribution or higher complexity of arrangement; a smaller fiber micro-difference index indicates that the fiber arrangement in the analysis sub-region is less different, and the fiber distribution tends to be uniform and has higher coherence.

[0088] Among them, the weights of pixel distribution parameters and neighborhood feature parameters determine the composition ratio of fiber micro-difference indicators. The weight setting needs to be adjusted according to specific analysis needs. If more attention is paid to grayscale changes, the weight of pixel distribution parameters should be increased; if more attention is paid to spatial correlation, the weight of neighborhood feature parameters should be increased to optimize the accuracy of evaluating fiber arrangement characteristics.

[0089] The complexity analysis of the distribution characteristics of fiber micro-difference indexes is carried out through image decomposition based on multi-fractal model to determine whether the randomness of the fiber arrangement pattern in the local area is normal, including:

[0090] The fiber differential indexes calculated in the analysis sub-area are distributed according to the spatial coordinates to generate a two-dimensional distribution matrix:

[0091] In the analysis sub-area, the fiber micro-difference index of each pixel point is spatially organized, and the corresponding two-dimensional distribution matrix is ​​generated according to the spatial coordinate distribution of the pixel; each element in the matrix corresponds to the fiber micro-difference index value of a pixel point, and the number of rows and columns of the matrix corresponds to the height and width of the image, respectively.

[0092] To ensure the accuracy of matrix generation, the calculation results of the fiber micro-difference index must cover all pixels in each sub-region, and the uncalculated areas are marked with missing values. In addition, the data interpolation method is used to fill in the empty values ​​caused by acquisition errors to ensure the integrity of the matrix.

[0093] Based on the multi-fractal model, the two-dimensional distribution matrix is ​​decomposed into multiple scale components:

[0094] The generated two-dimensional distribution matrix is ​​decomposed using a multifractal model to extract the features of the matrix at different scales. The specific operations include: decomposing the two-dimensional distribution matrix into several sub-matrices with different scale features, each sub-matrix corresponding to a decomposition scale; the decomposition process follows the hierarchical principle of the multifractal model, that is, gradually extracting small-scale features by reducing the size of the observation window while retaining the global macroscopic features.

[0095] Each decomposed sub-matrix retains the distribution characteristics of the original data at the corresponding scale and is used for subsequent complexity analysis.

[0096] The fractal dimension corresponding to each level component is calculated to quantify the complexity of fiber arrangement at different scales in the two-dimensional distribution matrix:

[0097] The calculation formula of fractal dimension is: ;in, Represents the fractal dimension, which is used to measure the complexity of data distribution at the current scale; The scale is The number of feature points covered by the sub-matrix; is the scale size of the current observation window.

[0098] The fractal dimension reflects the hierarchical complexity of fiber arrangement, and a higher fractal dimension indicates that the matrix data has more randomness and diversity.

[0099] According to the distribution law of fractal dimension at different scales, the complexity characteristic parameters that characterize the randomness of fiber arrangement are extracted:

[0100] Analyze the changing trend of fractal dimension at each scale, such as whether it shows monotonous decrease, oscillation or disordered distribution.

[0101] Parameters such as fractal spectrum width and fractal intensity are extracted as complexity characteristic parameters representing the complexity of fiber arrangement.

[0102] Among them, the fractal spectrum width refers to the range of variation of the fractal dimension at different scale components, which is used to quantify the diversity of data distribution. The fractal spectrum width is calculated by statistically analyzing the difference between the maximum and minimum values ​​of the fractal dimension. A larger fractal spectrum width indicates that the fiber arrangement has a wider range of multi-scale characteristics, and there may be complex random distribution and significant non-uniformity. A smaller fractal spectrum width indicates that the distribution characteristics of the fiber arrangement are more consistent and have higher uniformity. The fractal strength reflects the steepness of the change of the fractal dimension at different scale components, which is used to measure the complexity and randomness of the data distribution. The fractal strength is quantified by fitting the slope of the curve of fractal dimension and scale change. A higher fractal strength indicates that the fiber arrangement is more random and there may be significant microscopic changes. A lower fractal strength indicates that the fiber arrangement tends to be stable. The fractal spectrum width and fractal strength jointly characterize the complexity and distribution characteristics of the fiber arrangement.

[0103] Compare the complexity characteristic parameters with the preset standards to determine whether the randomness of the fiber arrangement pattern in the local area is normal:

[0104] Based on historical data and experimental samples, a reasonable range of fractal spectrum width and fractal intensity is established to form a preset standard for complexity characteristic parameters. The preset standard should take into account the differences in fiber arrangement in different regions to ensure that the standard has wide applicability.

[0105] For example, the reasonable range of fractal spectrum width and fractal intensity can be expressed by upper and lower limits respectively: fractal spectrum width is [Wmin, Wmax], and fractal intensity is [Smin, Smax].

[0106] If the fractal spectrum width is within a reasonable range, it means that the multi-scale characteristics of the fiber arrangement are normal. If it exceeds the range, there may be abnormal non-uniform distribution; if the fractal intensity is within a reasonable range, it means that the complexity and randomness of the fiber arrangement are normal. If it exceeds the range, there may be overly smooth or drastic random changes.

[0107] When both the fractal spectrum width and the fractal strength are within a reasonable range, the randomness of the fiber arrangement pattern in the local area is judged to be normal; if any parameter of the fractal spectrum width and the fractal strength is not within a reasonable range, the randomness of the fiber arrangement pattern in the local area is judged to be abnormal.

[0108] The nonlinear method based on phase space reconstruction is used to analyze the changing trend of fiber differential indicators at different time sequences to determine whether the deviation amplitude of fiber arrangement is normal, including:

[0109] Arrange the fiber differential indicators at each time point in order to form a fiber differential indicator time series:

[0110] The fiber micro-difference index values ​​corresponding to each time point are extracted and arranged in chronological order to generate a fiber micro-difference index time series.

[0111] The fiber micro-difference index time series is a sequence of data with time as the horizontal axis and fiber micro-difference index value as the vertical axis, which characterizes the dynamic characteristics of fiber arrangement changes at different time sequences; its length should cover the entire analysis period to ensure that it can reflect the complete dynamic characteristics of fiber arrangement. Any missing time point data needs to be supplemented by interpolation to ensure the integrity of the time series.

[0112] Phase space reconstruction is used to map the fiber differential index time series into a multidimensional state vector set:

[0113] According to the fiber differential index time series, the delayed coordinate method is used to map it into the high-dimensional phase space to generate a multi-dimensional state vector set.

[0114] Assume that the fiber differential index time series is expressed as , then the time point The state vector is represented as: ;in, Indicates time point The state vector of Indicates time point The fiber micro-difference index, is the time delay parameter (used to determine the delay interval of the state vector), is the embedding dimension (used to define the dimension size of the state vector).

[0115] is a specific state vector in the multidimensional state vector set, each It is part of a multidimensional state vector set and is used to describe the evolution law of time series in high-dimensional space.

[0116] Among them, the selection of time delay parameters is based on the mutual information method. By calculating the mutual information values ​​of the time series at different time intervals, the optimal delay value is determined. Specifically, as the time interval increases, the correlation between the sequences gradually weakens, and the mutual information value gradually decreases. By detecting the first local minimum point of the mutual information value, the corresponding time interval is selected as the time delay parameter. This method ensures that the mapped state vector is independent in the time dimension and avoids information redundancy caused by excessive correlation.

[0117] The determination of embedding dimension is based on the pseudo nearest neighbor method. By analyzing the embedding characteristics of time series in different dimensions, the optimal embedding dimension is found. The pseudo nearest neighbor method compares the points in the high-dimensional phase space with their neighboring relationships in the low-dimensional space. When the embedding dimension is insufficient, the neighboring points will become non-neighboring due to the folding phenomenon. The embedding dimension is gradually increased until the pseudo nearest neighbor ratio is less than the set threshold. At this time, the corresponding dimension is the embedding dimension.

[0118] Nonlinear analysis methods are applied to the multidimensional state vector set to extract the offset-related characteristic parameters, which are used to quantify the change law of fiber arrangement in the time dimension:

[0119] The Euclidean distance between the corresponding state vectors at adjacent time points is calculated as the local offset distance to quantify the instantaneous change amplitude of the fiber arrangement: ;in, Indicates time point The local offset distance, Indicates time point The state vector of Quantity, is the index of the component in the state vector (used to identify different dimensions of a high-dimensional state vector).

[0120] The larger the local offset distance, the greater the instantaneous change in fiber arrangement.

[0121] Compare the offset-related characteristic parameters with the preset reference range to determine whether the offset amplitude of the fiber arrangement is normal:

[0122] The preset reference range [Bmin, Bmax] is set based on a large amount of historical data and experimental sample statistics. The reasonable upper and lower limits are determined by analyzing the distribution law of the local offset distance of fiber arrangement under normal conditions. The specific range needs to take into account the dynamic change characteristics at different time points, regions or processing conditions to ensure adaptability and accuracy.

[0123] When the local offset distance is within the preset reference range [Bmin, Bmax], it means that the instantaneous change amplitude of the fiber arrangement is within a reasonable range and the dynamic characteristics are normal; if it exceeds the preset reference range [Bmin, Bmax], it means that the offset amplitude is abnormal, which may be due to excessive or insufficient dynamic changes, reflecting abnormal fiber arrangement or measurement errors.

[0124] When the local offset distance is within the preset reference range [Bmin, Bmax], the offset amplitude of the fiber arrangement is determined to be normal; otherwise, the offset amplitude of the fiber arrangement is determined to be abnormal.

[0125] Based on the randomness of the fiber arrangement pattern in the local area and the judgment result of whether the deviation amplitude of the fiber arrangement is normal, it is judged whether the stability of the fiber micro-difference index meets the standard, including:

[0126] The comprehensive evaluation based on the randomness of the fiber arrangement pattern and the offset amplitude can fully reflect the stability of the fiber micro-difference index. The randomness of the fiber arrangement pattern is used to evaluate the diversity and uniformity of the fiber distribution, ensuring that the fiber structure has reasonable complexity in the local area; and the offset amplitude reflects the stability of the fiber arrangement during dynamic changes, avoiding dynamic fluctuations that are too drastic or too gentle. Only when the randomness and offset amplitude are in a normal state can it be proved that the fiber arrangement has both reasonable structural characteristics and remains stable during dynamic changes, thereby ensuring the accuracy of the fiber micro-difference index. Such a comprehensive judgment logic can reduce the errors that may be caused by a single evaluation, comprehensively improve the reliability of the index, provide a scientific basis for subsequent analysis, and avoid inaccuracies caused by data anomalies.

[0127] Therefore, when the randomness of the fiber arrangement pattern in the local area is normal and the deviation amplitude of the fiber arrangement is normal, the stability of the fiber micro-difference index is judged to be up to standard; otherwise, the stability of the fiber micro-difference index is judged to be not up to standard.

[0128] When the stability of the fiber differential index meets the standard, the fiber differential index is compared with the fiber component reference data. After the fiber differential index meets the preset standard, the straw nutritional analysis data is determined, including:

[0129] Select the corresponding reference data from the preset fiber composition reference database according to the properties of the analysis sub-area:

[0130] Construction of fiber composition reference database: The fiber composition reference database is constructed based on experimental sample data and actual application data. It contains the fiber composition parameter ranges under different sources, processing methods and batches. Each data record includes the upper and lower limits of indicator values ​​such as cellulose content, hemicellulose content and lignin content.

[0131] The reference data was screened to ensure a good match based on the attributes of the sub-area of ​​analysis (e.g. fiber origin, processing method, and batch) using the following selection steps:

[0132] Exact match filtering: Prioritize reference entries that are completely consistent with the attributes of the analysis sub-area.

[0133] Close match filtering: When there are not enough exact matching entries, select the entry with the closest source attribute as a reference.

[0134] Priority screening: If multiple entries meet the criteria, select the best reference data based on source attribute priority (such as batch priority or processing method priority).

[0135] Compare the fiber difference index with the selected reference data item by item to determine whether it meets the preset standards:

[0136] The comparison items include nutritional properties such as cellulose content, hemicellulose content, and lignin content, each of which has reference upper and lower limits.

[0137] For each indicator, determine whether the calculated value falls within the range corresponding to the reference data: if the calculated value is within the upper and lower limits, mark it as a compliant item; if the calculated value exceeds the range, mark it as a non-compliant item and record the abnormal value.

[0138] For example, assuming the reference range of cellulose content is [30, 40], and the calculated value is 35, it is judged to be in compliance; if the calculated value is 45, it is judged to be non-compliant.

[0139] Compare all indicators in the sub-area one by one and generate a set of comparison result data (compliant / non-compliant).

[0140] When all comparison items between the fiber difference index and the reference data meet the preset standards, the straw nutrition analysis data is generated:

[0141] When the fiber micro-difference indicators of the sub-area meet the reference data range in all comparison items, it is judged to meet the preset standard; if any indicator does not meet the reference data range, it is judged to not meet the preset standard.

[0142] For sub-areas that do not meet the preset standards, it is necessary to record the abnormal items and their corresponding reference data and calculated values ​​to provide a basis for subsequent analysis.

[0143] For all analyzed sub-regions that meet the preset standards, their fiber differential index values ​​are integrated to generate comprehensive nutritional characteristic data of the sub-regions. The data integration includes the average value of cellulose content, the total amount of hemicellulose content and the distribution of lignin content.

[0144] Based on the combined data of all analyzed sub-regions, the overall nutritional analysis results for the entire sample were calculated. The global analysis data included the average cellulose content, average hemicellulose content, and lignin distribution ratio and overall nutritional characteristics.

[0145] The generated straw nutritional analysis data is stored in the database for feed formula adjustment or production process optimization. If a sub-area does not meet the preset standards, the corresponding area will be marked as an abnormal area to prompt further analysis and adjustment.

[0146] Example 2: The difference between Example 2 of the present invention and Example 1 is that this example introduces a straw feed nutrition analysis system based on image recognition.

[0147] Figure 2 A structural schematic diagram of the straw feed nutrition analysis system based on image recognition of the present invention is given. The straw feed nutrition analysis system based on image recognition includes an optical imaging acquisition module, a fiber differential extraction module, a random complex analysis module, an offset amplitude analysis module, a stable comprehensive judgment module and a nutrition data generation module.

[0148] Optical imaging acquisition module: multi-directional optical imaging acquisition is performed on the straw surface to obtain straw image data, and the straw image data is divided into multiple analysis sub-areas.

[0149] Fiber micro-difference extraction module: extracts fiber arrangement information of the analysis sub-area based on morphological analysis to enhance fiber texture features, and extracts fiber micro-difference indicators based on pixel distribution and neighborhood relationships.

[0150] Random complex analysis module: Through image decomposition based on multi-fractal model, the distribution characteristics of fiber micro-difference indicators are analyzed for complexity to determine whether the randomness of the fiber arrangement pattern in the local area is normal.

[0151] Offset amplitude analysis module: The nonlinear method based on phase space reconstruction is used to analyze the changing trend of fiber differential indicators at different time sequences to determine whether the offset amplitude of the fiber arrangement is normal.

[0152] Stability comprehensive judgment module: Based on the randomness of the fiber arrangement pattern in the local area and whether the deviation amplitude of the fiber arrangement is normal, it is judged whether the stability of the fiber micro-difference index meets the standard.

[0153] Nutritional data generation module: When the stability of the fiber differential index meets the standard, the fiber differential index is compared with the fiber composition reference data. After the fiber differential index meets the preset standard, the straw nutritional analysis data is determined.

[0154] 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 and thresholds in the formula are set by technicians in this field according to actual conditions.

[0155] 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. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions may be stored in a computer-readable storage medium, or may be transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions may be transmitted from one website, computer, server or data center to another website, computer, server or data center by wired (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium may be any available medium that a computer can access or a data storage device such as a server or data center that contains one or more available media sets. The available medium may be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium. The semiconductor medium may be a solid-state hard disk.

[0156] Those of ordinary skill in the art will appreciate that the modules and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0157] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and modules described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0158] In the several embodiments provided in the present application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the modules is only a logical function division. There may be other division methods in actual implementation, such as multiple modules or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or modules, which can be electrical, mechanical or other forms.

[0159] The modules described as separate components may or may not be physically separated, and the components shown as modules may or may not be physical modules, and may be located in one place or distributed on multiple network modules. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0160] In addition, each functional module in each embodiment of the present application may be integrated into one processing module, or each module may exist physically separately, or two or more modules may be integrated into one module.

[0161] If the functions are implemented in the form of software function modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application can be essentially or partly embodied in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods described in the various embodiments of the present application. The aforementioned storage media include: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk, and other media that can store program codes.

[0162] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present 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. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

[0163] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A straw feed nutrition analysis method based on image recognition, characterized in that: The steps include: Perform multi-directional optical imaging on the straw surface to obtain straw image data, and divide the straw image data into multiple analysis sub-areas; The fiber arrangement information of the sub-region is extracted based on morphological analysis to enhance the fiber texture characteristics, and the fiber micro-difference index is extracted based on pixel distribution and neighborhood relationship; The complexity analysis of the distribution characteristics of fiber micro-difference indexes is carried out through image decomposition based on multi-fractal model to determine whether the randomness of fiber arrangement pattern in local area is normal. The variation trend of fiber micro-difference index at different time sequences is analyzed by nonlinear method based on phase space reconstruction to judge whether the deviation amplitude of fiber arrangement is normal. Based on the randomness of the fiber arrangement pattern in the local area and the judgment result of whether the deviation amplitude of the fiber arrangement is normal, it is judged whether the stability of the fiber micro-difference index meets the standard; When the stability of the fiber micro-difference index meets the standard, the fiber micro-difference index is compared with the fiber composition reference data. After the fiber micro-difference index meets the preset standard, the straw nutritional analysis data is determined.

2. The straw feed nutrition analysis method based on image recognition according to claim 1, characterized in that: Multi-directional optical imaging is performed on the straw surface to obtain straw image data, and the straw image data is divided into multiple analysis sub-areas, including: A plurality of optical imaging devices are arranged on the surface of the straw and a plurality of light sources are configured, and the image data of the straw in multiple directions are recorded by adjusting the focal length and shooting parameters of the optical imaging devices; The multi-directional straw image data are stored and preprocessed to form a preprocessed image set, and the preprocessed image set is decomposed into multiple analysis sub-areas according to a preset division criterion.

3. The straw feed nutrition analysis method based on image recognition according to claim 2, characterized in that: The fiber arrangement information of the sub-region is extracted based on morphological analysis to enhance the fiber texture characteristics, and the fiber micro-difference index is extracted based on pixel distribution and neighborhood relationship, including: Morphological analysis is applied to the straw image data in the analysis sub-area, and fiber texture features are enhanced by selecting multi-scale structural elements, which are used to adapt to fiber textures of different thicknesses and directions. The fiber arrangement information in the analysis sub-area is quantitatively analyzed based on the pixel distribution parameters, which are obtained by calculating the gray value variation characteristics of the local area; The neighborhood relationship of fiber arrangement information is analyzed, and the neighborhood characteristic parameters of fiber arrangement are extracted based on the spatial correlation between local pixels and their surrounding pixels. The fiber micro-difference index is extracted by combining pixel distribution parameters with neighborhood characteristic parameters. The fiber micro-difference index is used to characterize the microscopic difference characteristics of fiber arrangement.

4. The straw feed nutrition analysis method based on image recognition according to claim 3, characterized in that: The complexity analysis of the distribution characteristics of fiber micro-difference indexes is carried out through image decomposition based on multi-fractal model to determine whether the randomness of the fiber arrangement pattern in the local area is normal, including: Generate a two-dimensional distribution matrix by distributing the fiber differential indexes calculated in the analysis sub-area according to the spatial coordinates; Decomposing the two-dimensional distribution matrix into multiple scale components based on the multi-fractal model; The fractal dimension corresponding to each level of component is calculated to quantify the complexity of fiber arrangement at different scales in the two-dimensional distribution matrix; According to the distribution law of fractal dimension at different scales, the complexity characteristic parameters representing the randomness of fiber arrangement are extracted. The complexity characteristic parameters include fractal spectrum width and fractal intensity. The complexity characteristic parameters are compared with the preset standards to determine whether the randomness of the fiber arrangement pattern in the local area is normal.

5. The straw feed nutrition analysis method based on image recognition according to claim 4, characterized in that: The nonlinear method based on phase space reconstruction is used to analyze the changing trend of fiber differential indicators at different time sequences to determine whether the deviation amplitude of fiber arrangement is normal, including: The fiber micro-difference indexes at each time point are arranged in order to form a fiber micro-difference index time series; The fiber differential index time series is mapped into a multi-dimensional state vector set by phase space reconstruction; Applying nonlinear analysis methods to the multidimensional state vector set, the offset-related characteristic parameters are extracted. The offset-related characteristic parameters are used to quantify the variation of fiber arrangement in the time dimension. The deviation-related characteristic parameters are compared with the preset reference range to determine whether the deviation amplitude of the fiber arrangement is normal.

6. The straw feed nutrition analysis method based on image recognition according to claim 5, characterized in that: The nonlinear analysis method is applied to the multidimensional state vector set to extract the offset-related characteristic parameters, specifically: Calculate the Euclidean distance of the corresponding state vectors at adjacent time points as the local offset distance: ;in, Indicates time point The local offset distance, Indicates time point The state vector of Quantity, is the index of the component in the state vector, is the embedding dimension.

7. The straw feed nutrition analysis method based on image recognition according to claim 6, characterized in that: Based on the randomness of the fiber arrangement pattern in the local area and the judgment result of whether the deviation amplitude of the fiber arrangement is normal, it is judged whether the stability of the fiber micro-difference index meets the standard, including: When the randomness of the fiber arrangement pattern in the local area is normal and the deviation amplitude of the fiber arrangement is normal, it is determined that the stability of the fiber micro-difference index meets the standard; otherwise, it is determined that the stability of the fiber micro-difference index does not meet the standard.

8. The straw feed nutrition analysis method based on image recognition according to claim 7, characterized in that: When the stability of the fiber differential index meets the standard, the fiber differential index is compared with the fiber component reference data. After the fiber differential index meets the preset standard, the straw nutritional analysis data is determined, including: Selecting corresponding reference data from a preset fiber component reference database according to the attributes of the analysis sub-region; Compare the fiber difference index with the selected reference data item by item to determine whether it meets the preset standards; When all comparison items between the fiber differential index and the reference data meet the preset standards, the straw nutritional analysis data is generated.

9. A straw feed nutrition analysis system based on image recognition, used to implement the straw feed nutrition analysis method based on image recognition according to any one of claims 1 to 8, characterized in that: It includes optical imaging acquisition module, fiber micro-difference extraction module, random complex analysis module, offset amplitude analysis module, stable comprehensive judgment module and nutrition data generation module; Optical imaging acquisition module: performs multi-directional optical imaging acquisition on the straw surface to obtain straw image data, and divides the straw image data into multiple analysis sub-areas; Fiber micro-difference extraction module: extracts fiber arrangement information of the analysis sub-region based on morphological analysis to enhance fiber texture features, and extracts fiber micro-difference indicators based on pixel distribution and neighborhood relationship; Random Complex Analysis Module: Through image decomposition based on multi-fractal model, the distribution characteristics of fiber micro-difference index are analyzed for complexity, and the randomness of fiber arrangement pattern in the local area is judged to be normal; Offset amplitude analysis module: Analyze the variation trend of fiber differential index at different time sequences through nonlinear method based on phase space reconstruction to determine whether the offset amplitude of fiber arrangement is normal; Stability comprehensive judgment module: Based on the randomness of the fiber arrangement pattern in the local area and whether the deviation amplitude of the fiber arrangement is normal, it is judged whether the stability of the fiber micro-difference index meets the standard; Nutritional data generation module: When the stability of the fiber differential index meets the standard, the fiber differential index is compared with the fiber composition reference data. After the fiber differential index meets the preset standard, the straw nutritional analysis data is determined.

Images