A method and system for monitoring the state of seedling growth in a root enhancing container

Abstract

The present application relates to the technical field of image data processing, and more particularly to a kind of root increasing container seedling state monitoring method and system, the method includes steps: collecting root increasing container side wall image and conversion to Lab space;Calculate global color statistical characteristics to determine the initial feature response value of pixel point;Determine local contrast weight based on the difference of brightness and neighborhood mean, and obtain chrominance effective coefficient by using chrominance correction;Combining initial feature response value and chrominance effective coefficient obtains final feature response value, and evaluates seedling state accordingly.The present application effectively suppresses container background drift and inorganic substrate particle interference by fusing global color difference, local contrast and biological chroma characteristics, and improves the accuracy of root system penetration monitoring.

Description

Technical Field

[0001] This invention relates to the field of image data processing technology, and in particular to a method and system for monitoring the status of root-promoting container seedling cultivation. Background Technology

[0002] Rooting containers, utilizing their sidewall ventilation holes and air trimming principles, limit taproot coiling and promote fibrous root growth, making them a key material for improving seedling transplant survival rates. In large-scale mobile nurseries, real-time monitoring of root conditions at the ventilation holes is crucial. This not only reflects water management levels but also serves as a basis for assessing the failure of root control mechanisms. If abnormal root penetration goes undetected, it will lead to drying and rotting, reducing seedling quality. Therefore, accurate root monitoring is of significant practical importance for optimizing irrigation strategies and ensuring seedling quality.

[0003] Currently, machine vision-based image processing technology has been applied to seedling monitoring. Among them, the Frequency Tuning (FT) algorithm is often used for initial target extraction due to its simple calculation and fast response. This algorithm utilizes color space transformation and defines areas with high color deviation in the image by calculating the Euclidean distance between the pixel color and the mean of the entire image. In the case of a simple background, this algorithm can quickly locate objects with large color differences and is often used to analyze sidewall images to assist in identifying root targets at the aeration holes.

[0004] However, applying the Fourier Transform (FT) algorithm to unstructured seedling nursery sites has limitations: real-world scenarios contain complex elements such as supports and the ground, causing a shift in the global average color value of the image. In this situation, the dark black container itself, due to its large difference from the shifted global mean, is often misjudged as the foreground subject, thus obscuring the tiny, light-colored roots that truly need to be monitored. Furthermore, the substrate often contains white inorganic particles such as perlite, whose visual characteristics are similar to newly formed white roots. Relying solely on global color distance is insufficient to distinguish such bright interference, leading to the misidentification of inorganic particles as roots, resulting in false detections and biased monitoring results. Summary of the Invention

[0005] To address the technical problems of dark container bodies obscuring root targets and white inorganic matrix particles confusing visual features of roots due to global color mean shift in complex background environments, this invention provides solutions in the following aspects.

[0006] In a first aspect, the present invention provides a method for monitoring the status of root-promoting container seedlings, the method comprising the steps of:

[0007] Acquire sidewall images of the root-promoting container and convert them to the Lab color space to obtain a Lab image. Calculate the global color statistical features of the Lab image. For each pixel in the Lab image, determine the initial feature response value of the pixel based on the color difference between its color vector and the global color statistical features. Determine the local contrast weight of the pixel based on the difference between its luminance value and the average luminance value within its neighborhood window, and the average luminance value of the neighborhood window. Calculate the chromaticity of the pixel and use it to correct the local contrast weight to obtain the effective chromaticity coefficient of the pixel. Multiply the initial feature response value by the effective chromaticity coefficient to obtain the final feature response value of the pixel. Determine whether roots have penetrated the root system based on the final feature response value to assess the seedling status of the root-promoting container.

[0008] This invention acquires images of the sidewalls of root-promoting containers and converts them to the Lab color space. It then uses global color statistical features to calculate color difference metrics to obtain initial feature response values, thereby providing a preliminary assessment of potential target areas. Considering the complex background of the seedling nursery environment, relying solely on global color features is easily affected by container color drift or non-root impurities. Therefore, this invention further introduces local contrast weights. By analyzing the difference between pixel brightness values ​​and the average brightness within their neighborhood windows, it highlights suspected targets that exhibit high brightness relative to the black container background. Simultaneously, it uses chromaticity calculation to adjust the local contrast weights. Leveraging the physical difference between the biological color of roots and the achromatic properties of matrix particles such as perlite, it reduces interference from inorganic particles. Finally, the initial feature response value is multiplied by the chromaticity effective coefficient to obtain the final feature response value. This multi-feature fusion processing method can retain weak root signals while suppressing background noise, improving the accuracy of monitoring results for root-promoting container seedling status.

[0009] Preferably, determining the local contrast weight of a pixel based on the difference between the brightness value of each pixel and the average brightness value within its neighborhood window, and the average brightness value of the neighborhood window, includes: calculating the ratio of the difference to the average brightness value of the neighborhood window to obtain a relative contrast coefficient; determining a brightness modulation coefficient based on the average brightness value of the neighborhood window, wherein the brightness modulation coefficient is negatively correlated with the average brightness value of the neighborhood window; and calculating the local contrast weight based on the product of the relative contrast coefficient and the brightness modulation coefficient.

[0010] Preferably, the local contrast weights satisfy the following relationship:

[0011] ;

[0012] in, It is the first Local contrast weights for each pixel; It is the first The brightness value of each pixel; It is the first The average brightness of the neighborhood window of each pixel; This is the preset brightness reference value; It is a preset micro value; Basic weighting coefficients; It is a natural exponential function.

[0013] This invention employs a specific relational formula containing a natural exponential function to calculate local contrast weights. The normalized difference between the brightness value and the mean of the neighborhood is used to characterize the relative prominence of the target, and the background brightness is evaluated through the natural exponential term. When the mean brightness of the neighborhood window approaches the preset brightness reference value, i.e., the brightness of the container itself, the function output value increases. This calculation method combines the screening of the high-brightness features of the root system with the verification of the dark background of the container. Through mathematical operations, the visual contrast features are transformed into numerical weights, which helps to stably extract suspected root system pixels on the surface of the container with growing roots under complex lighting conditions.

[0014] Preferably, the step of correcting the local contrast weight using the chromaticity to obtain the effective chromaticity coefficient of the pixel includes: determining a chromaticity adjustment coefficient based on the chromaticity, wherein the chromaticity adjustment coefficient is positively correlated with the chromaticity; and multiplying the local contrast weight by the chromaticity adjustment coefficient to obtain the effective chromaticity coefficient.

[0015] Preferably, the effective chromaticity coefficient satisfies the following relationship:

[0016] ;

[0017] in, It is the first The effective chromaticity coefficient of each pixel; It is the first Local contrast weights for each pixel; , They are the first The red-green and yellow-blue components of each pixel in the Lab color space; It is a natural exponential function.

[0018] This invention uses a natural exponential relationship containing the moduli of red-green and yellow-blue components to calculate the effective chromaticity coefficient. The square root of the sum of the squares of the red-green and yellow-blue components is used to characterize the chromaticity saturation of a pixel. When the saturation is high, the negative exponential term approaches zero, allowing the weight to be retained. When the saturation approaches zero, the entire coefficient decays. This calculation method can transform the abstract difference in color saturation into numerical modulation of the weight, achieving mathematical suppression of bright inorganic interference and ensuring that the final feature response value reflects the biologically active root target more.

[0019] Preferably, the global color statistical feature is the global mean vector of the Lab image, which is obtained by: extracting the luminance component, red-green component, and yellow-blue component of each pixel in the Lab image in the Lab color space; calculating the arithmetic mean of the luminance component, red-green component, and yellow-blue component of all pixels in the Lab image to obtain the mean of the luminance component, the mean of the red-green component, and the mean of the yellow-blue component; and constructing the global mean vector from the mean of the luminance component, the mean of the red-green component, and the mean of the yellow-blue component.

[0020] This invention extracts the brightness component, red-green component, and yellow-blue component of each pixel in the Lab image, and calculates the arithmetic mean of each component to form a global mean vector. This establishes a statistical benchmark that reflects the overall color tone of the current sidewall image. This feature acquisition method based on full-image statistics can quickly capture the overall illumination and color level of the seedling site environment, providing a reliable reference for subsequent calculation of the color abruptness of each pixel. It also helps to maintain the relative stability of the monitoring benchmark at inspection times with different light intensities.

[0021] Preferably, determining the initial feature response value of the pixel based on the color difference measure between its color vector and the global color statistical features includes: obtaining the initial feature response value of each pixel based on the square of the Euclidean distance between the color vector of each pixel in the Lab image and the global mean vector.

[0022] This invention obtains the initial feature response value based on the square of the Euclidean distance between the color vector of each pixel and the global mean vector. The Euclidean distance is used to measure the degree of deviation of a pixel from the background reference in the color space. The larger the distance, the more visually obvious the point is. This calculation method can quickly locate areas with large color differences in the image and initially screen out potential root targets from a large area of ​​uniform container background, providing a primary response map containing rich information for subsequent refined feature denoising.

[0023] Preferably, the step of determining whether roots have emerged based on the final feature response value to assess the seedling status of the root-promoting container includes: constructing a grayscale feature response map based on the final feature response values ​​of all pixels; performing binarization segmentation and morphological processing on the grayscale feature response map to extract connected regions; determining the effective root region based on the geometric morphological features of the connected regions; and assessing the seedling status of the root-promoting container based on the area ratio of the effective root region.

[0024] Preferably, the assessment of the seedling status of the root-promoting container based on the area ratio of the effective root system region includes: counting the total area of ​​all pixels in the Lab image that are determined to be effective root systems; calculating the ratio of the total area to the theoretical total area of ​​the vents in the root-promoting container; and generating and outputting a warning signal in response to the ratio exceeding a preset warning threshold.

[0025] In a second aspect, the present invention provides a root-promoting container seedling status monitoring system, which includes a memory and a processor. The memory stores computer program instructions, and when the computer program instructions are executed by the processor, a root-promoting container seedling status monitoring method of the first aspect of the present invention is implemented.

[0026] By adopting the above technical solution, a computer program for monitoring the status of root-promoting container seedling cultivation according to the first aspect of the present invention is generated and stored in a memory so that it can be loaded and executed by a processor. A terminal device can be made based on the memory and the processor for convenient use.

[0027] The beneficial effects of this invention are as follows: This invention utilizes a multi-feature fusion strategy in the Lab color space for monitoring. By combining global color statistical features, local brightness contrast, and chromaticity saturation information, it overcomes the problem of single color features being easily affected by container color drift and interference from white matrix particles in complex backgrounds, thus improving the accuracy of identifying tiny root targets in unstructured seedling cultivation sites. This invention introduces an adaptive weighting mechanism, which enhances the response of bright targets on dark container backgrounds by analyzing the brightness difference between pixels and neighboring windows and background blackness. It also uses biochromatic features to suppress achromatic inorganic impurities, thereby achieving robust extraction of specific biological targets without relying on a large number of sample training. This invention establishes a complete monitoring chain from pixel-level response to geometric morphology screening to area ratio assessment. It not only focuses on the existence of roots but also combines morphological features to remove interference and finally outputs quantitative state assessment indicators, which can provide timely feedback on the root control effect and water status of root-promoting containers, helping to improve the level of refined management in large-scale seedling cultivation. Attached Figure Description

[0028] Figure 1 A flowchart illustrating a method for monitoring the status of root-promoting container seedlings provided in an embodiment of the present invention;

[0029] Figure 2 This is a schematic diagram of the sidewall image of the root-addition container acquired according to an embodiment of the present invention;

[0030] Figure 3 A visual schematic diagram of the initial feature response value matrix provided in an embodiment of the present invention;

[0031] Figure 4 This is a comparative diagram of the root penetration status assessment results provided in an embodiment of the present invention;

[0032] Figure 5 This is a structural block diagram of a root-promoting container seedling status monitoring system provided in an embodiment of the present invention. Detailed Implementation

[0033] The first aspect of this invention provides a method for monitoring the status of root-promoting container seedling cultivation, such as... Figure 1 As shown, the method includes steps S100-S400:

[0034] Step S100: Acquire the sidewall image of the root-adding container, convert the sidewall image to the Lab color space, and obtain the Lab image.

[0035] It should be noted that high-quality source images are a prerequisite for accurate monitoring. Due to the variable lighting conditions and presence of substrate dust in the seedling cultivation environment, direct analysis using raw RGB images is easily affected by uneven lighting and high-frequency noise. Furthermore, the strong correlation between channels in the RGB color space makes it difficult to separate luminance and color information for independent physical feature analysis. Therefore, this step aims to acquire low-noise sidewall images and convert them to the Lab color space, which better matches human visual perception, for subsequent independent analysis of luminance and chromaticity features.

[0036] Specifically, using an image acquisition device mounted on a mobile inspection robot, the shooting angle was adjusted to be as perpendicular as possible to the sidewall of the root-addition container to reduce perspective distortion, and RGB images including the vent area were acquired. Subsequently, a Gaussian filtering algorithm was used to denoise the acquired raw RGB images, eliminating sensor thermal noise and salt-and-pepper noise caused by on-site dust. After processing, the RGB images were converted to the Lab color space, separating the luminance component, red-green component, and yellow-blue component to obtain the Lab image. In this process, the luminance component was used for subsequent contrast analysis, and the color component was used for biometric analysis.

[0037] like Figure 2 As shown, this image is a schematic diagram of the side wall of the rooting container. The image includes not only the dark black rooting container body and the white fibrous roots growing at the vent holes, but also white perlite matrix particles and greenhouse supports and messy lighting in the background.

[0038] At this point, the Lab image of the root-adding container has been obtained.

[0039] Step S200: Calculate the global color statistical features of the Lab image; for each pixel in the Lab image, determine the initial feature response value of the pixel based on the color difference measure between its color vector and the global color statistical features.

[0040] It should be noted that in the context of root-promoting container seedling cultivation, the most fundamental difference between the monitoring target (root system) and the background (container and environment) lies in the statistical distribution difference of their colors. To initially locate potential targets from complex backgrounds, this invention preferably employs the Frequency-Tune Algorithm (FT). The FT algorithm is a method that defines regional importance by analyzing the frequency domain features of an image. It has the advantages of simple computational logic, fast response speed, and the ability to output full-resolution results, making it suitable for industrial inspection scenarios with high real-time requirements. Based on this, this step utilizes the core idea of ​​the FT algorithm to evaluate the degree of deviation of each pixel from the average color level of the Lab image. The greater the Euclidean distance between the color vector of a pixel and the global mean vector, the more visually abrupt the point appears, and the greater its initial response strength as a potential target.

[0041] Specifically, each pixel in the augmented container Lab image is traversed, and the luminance, red-green, and yellow-blue components of the pixel in the Lab color space are extracted. The arithmetic mean of the luminance, red-green, and yellow-blue components of all pixels in the Lab image is calculated to obtain the mean values ​​of the luminance, red-green, and yellow-blue components. These mean values ​​constitute the global mean vector. For each pixel in the Lab image, the Euclidean distance between the pixel's color component vector and the global mean vector is calculated, and the square of the Euclidean distance is defined as the initial feature response value of the pixel.

[0042] Based on the above logic, the initial feature response value satisfies the following relationship:

[0043] ;

[0044] in, It is the first Initial feature response values ​​of each pixel; , , They are the first The brightness component, red-green component, and yellow-blue component of each pixel in the Lab color space; , , These are all pixels in the Lab image. , , The average value across the three channels.

[0045] In this formula, each term on the right-hand side represents the nth term. The square of the geometric distance between the color vector of the i-th pixel and the global mean vector; the larger this value, the stronger the geometric distance. The more a pixel's color deviates from the overall tone of the Lab image, i.e., the background color, the larger its initial feature response value.

[0046] like Figure 3 The image shows a visualization of the initial feature response value matrix. The bright, warm areas represent higher initial feature response values, while the dark, cool areas represent lower response values. As can be seen from the image, due to the use of the global mean to calculate the square of the Euclidean distance, the bright areas in the image background, such as the sky or the reflective greenhouse film, and the white perlite particles at the vents, all exhibit high response intensity. The dark container background and the bright non-root objects obscure the true root targets.

[0047] At this point, the initial feature response values ​​of each pixel have been obtained.

[0048] Step S300: Determine the local contrast weight of the pixel based on the difference between the brightness value of each pixel and the average brightness value in its neighborhood window, and the average brightness value of the neighborhood window; calculate the chromaticity of the pixel, and use the chromaticity to correct the local contrast weight to obtain the effective chromaticity coefficient of the pixel.

[0049] It should be noted that the complex background of the seedling site can cause the overall image mean to drift, resulting in abnormally high response values ​​for the dark black container itself due to its significant deviation from the global mean, thus obscuring the actual root system. Simultaneously, the white perlite particles mixed in the substrate have high brightness, making them difficult to distinguish from the roots in terms of brightness characteristics. Therefore, this step introduces local contrast weights and effective chromaticity coefficients to accurately describe the local high-contrast characteristic of white roots on black containers to eliminate container interference, and to eliminate perlite interference by recognizing the biological color of the roots and the achromatic nature of perlite.

[0050] Regarding the construction of local contrast weights, it's important to note that in the context of root-growth container seedling cultivation, the typical physical characteristics of root targets are as finely highlighted objects attached to the dark container surface. That is, a true root pixel should have significantly higher brightness than its surrounding neighborhood, while its surrounding neighborhood should exhibit the low-brightness characteristics characteristic of root-growth containers. Therefore, it's necessary to introduce relative brightness analysis of the local neighborhood to construct an index that can simultaneously characterize its own brightness and the deep black background, in order to selectively identify effective targets.

[0051] Specifically, first, iterate through each pixel in the Lab image and define the surrounding pixels of that pixel. The pixel region is a neighborhood window. The brightness values ​​of all pixels within the neighborhood window are counted and their arithmetic mean is calculated, which is denoted as the local brightness mean.

[0052] As a preferred embodiment, the neighborhood window size can be set to... In terms of pixels, considering that the primary roots at the vent holes of the root-addition container usually appear as fine linear structures, their diameter typically occupies 1 to 3 pixels in width in Lab images, The pixel neighborhood ensures that the window covers the root system and its adjacent container background, thus accurately calculating the brightness contrast between the background and the target, while also avoiding brightness interference from distant non-container backgrounds due to an excessively large window. Implementers can select appropriate parameters based on the actual shooting distance and the thickness of the root system image caused by the lens focal length; for example, for close-up macro photography of robust root systems, different parameters can be used. Pixels are selected for shooting delicate root systems at a distance. Adjust the neighborhood size of pixels.

[0053] Next, the local contrast weights are calculated. It should be noted that, in order to convert the local brightness mean into a calculable mathematical quantity, the constructed mathematical relationship satisfies the following rules: when a pixel is brighter than the background, the weight value increases; and when the background is darker, that is, closer to the brightness of the container itself, the gain effect of the weight is stronger, thereby achieving targeted enhancement of specific targets.

[0054] Based on the above logic, the local contrast weights satisfy the following relationship:

[0055] ;

[0056] in, It is the first Local contrast weights for each pixel; It is the first The brightness value of each pixel; It is the first The average brightness of the neighborhood window of each pixel; This is the preset brightness reference value; It is a preset microvalue used to prevent If the value is 0, it can be set to 0.001; Basic weighting coefficients; It is a natural exponential function.

[0057] In this formula, Measured the first The relative brightness difference of each pixel relative to its local background. The larger this value is, the brighter the foreground is relative to the background, which is consistent with the highlight characteristics of the root system. The reliability of a local background matching the characteristics of a black container was measured by the local brightness mean. The lower the value, the closer it is to black, the closer the value of this index term is to 1, thus amplifying the difference in relative brightness; conversely, if the background is brighter, this term is closer to 0, thus suppressing the response of non-container areas.

[0058] It should be noted that the preset brightness reference value and basic weight coefficient The value needs to be set according to the actual working conditions: for scenarios where the root-adding container material is matte dark black and the surface is relatively clean, It can be set to a lower value, such as 20, to strictly limit the range of background blackness, ensuring that only bright targets against a very dark background are enhanced; for scenes where the container surface has slight reflections or a small amount of dust, This can be increased appropriately, such as setting it to 40, to improve tolerance for non-ideal black backgrounds. (Regarding the base weight coefficient...) If the seedling environment has sufficient light, the contrast between the root system and the container will be extremely high. It can be set to a lower value, such as 0.1, to maximize the suppression of non-target areas; if the ambient light is dim and the overall contrast decreases, The value can be appropriately increased, such as setting it to 0.3, to retain the weak target signal and prevent excessive suppression. In this embodiment, the general seedling environment is taken into account. The preferred value is set to 20. The preferred value is 0.2.

[0059] Regarding the construction of effective chromaticity coefficients, it should be noted that both perlite and roots appear as bright objects against a dark background. However, as active plant tissue, roots typically possess a faint biological hue, such as a creamy yellow, which manifests as a non-zero chromaticity component in the Lab color space. Perlite, on the other hand, is an inorganic mineral, exhibiting achromatic pure white or grayish-white characteristics, with its chromaticity component approaching zero. Based on this difference in physical properties, this invention introduces chromaticity saturation analysis to further identify targets with biological attributes from the bright candidate regions.

[0060] It should be further noted that, in order to use chromaticity differences to eliminate inorganic particle interference, the construction of the effective chromaticity coefficients should follow the following principles: when the chromaticity saturation of a pixel is low, it is more likely that the pixel belongs to perlite, and its local contrast weight should be reduced; while when the chromaticity saturation reaches a certain level, it is determined that it has biological characteristics, and its weight should be retained.

[0061] Based on the above logic, the effective chromaticity coefficients satisfy the following relationship:

[0062] ;

[0063] in, It is the first The effective chromaticity coefficient of each pixel; It is the first Local contrast weights for each pixel; , They are the first The red-green and yellow-blue components of each pixel in the Lab color space; It is a natural exponential function.

[0064] In this formula, Characterizing the first The magnitude of the nth pixel in the chromaticity plane of the Lab color space, i.e., chromaticity. When this value is large, it indicates that the nth pixel... Each pixel exhibits distinct biological coloration, at which point the negative exponential function term... Approaching 0, making The negative exponential function term approaches 1, thus fully preserving the local contrast weights. The value of ; conversely, when the value approaches 0, it indicates that the pixel exhibits achromatic characteristics, consistent with the characteristics of perlite. At this time, the negative exponential function term approaches 1, making The negative exponential function term approaches 0, thus... The value is attenuated to negligible levels, thus effectively filtering out inorganic interference in high brightness.

[0065] At this point, the effective chromaticity coefficients of each pixel have been obtained.

[0066] Step S400: Multiply the initial feature response value by the effective chromaticity coefficient to obtain the final feature response value of the pixel; determine whether there are roots penetrating based on the final feature response value to evaluate the seedling status of the rooting container.

[0067] It should be noted that, considering that the initial feature response value only reflects the degree of abruptness of the global color and has not yet eliminated the interference of the container and perlite, this step uses the effective chromaticity coefficient as a gain factor to apply to the initial feature response value to generate a final monitoring map that can accurately characterize the root system target. Furthermore, morphological and geometric feature analysis techniques are used to transform the pixel-level response values ​​into a quantitative assessment conclusion on the root penetration status.

[0068] Specifically, firstly, the initial feature response value of each pixel is multiplied and modulated using the effective chromaticity coefficient of each pixel to calculate the final feature response value.

[0069] Based on the above logic, the final feature response value satisfies the following relationship:

[0070] ;

[0071] in, It is the first The final feature response value of each pixel; It is the first Initial feature response values ​​of each pixel; It is the first The effective chromaticity coefficient of each pixel.

[0072] In this formula, the multiplication operation implements a logical AND operation for multiple features. Its significance lies in the fact that only when a pixel simultaneously satisfies the global color abruptness... Larger, and consistent with the characteristics of root system organisms and localized light, i.e. When it is large, its final characteristic response value Only then will high values ​​be retained; if any condition is not met, the final characteristic response value will be suppressed, thereby eliminating background noise.

[0073] Next, traverse every pixel in the Lab image, calculate the final feature response value of all pixels, arrange these values ​​according to the original spatial position of the pixels, and construct a final feature response value matrix with the same size as the Lab image.

[0074] Then, image segmentation and morphological processing are performed on the final feature response value matrix. It should be noted that the final feature response values ​​are continuous numerical data and need to be mapped to a standard grayscale space for image segmentation. Furthermore, for the monitoring scenario at the vent holes of the root-growing container, the root system targets typically occupy a very small proportion in the Lab image and their feature signals are relatively weak. Relying on a fixed absolute threshold can easily lead to segmentation failure due to lighting fluctuations or sample differences. Therefore, an adaptive thresholding strategy based on the current image statistical features is required. Simultaneously, considering the unavoidable high-frequency noise interference such as dust reflection in the field environment, a morphological filtering mechanism needs to be introduced after target extraction to filter out isolated noise points while preserving the connectivity and integrity of the linear root system structure to the greatest extent possible.

[0075] Specifically, the final feature response value matrix is ​​normalized, linearly mapping all values ​​within the matrix to a grayscale space of 0 to 255, generating a grayscale feature response map. The maximum grayscale value in the grayscale feature response map is counted, and a segmentation threshold is set to 0.8 times this maximum grayscale value. The grayscale value of each pixel in the grayscale feature response map is compared with the segmentation threshold; pixels with grayscale values ​​greater than the segmentation threshold are marked as foreground objects, and pixels with grayscale values ​​less than or equal to the segmentation threshold are marked as background, thus obtaining a binarized image. Finally, morphological opening operations are performed on the binarized image to obtain a morphologically processed binarized image. This involves first performing an erosion operation to eliminate isolated small noise points such as dust reflections, followed by a dilation operation to restore the shape of the main connected regions.

[0076] Next, geometric feature selection and status warning are performed on the binarized image after morphological processing.

[0077] Specifically, connected component analysis is performed on the binarized image after morphological processing, and the geometric attributes of each connected region are statistically analyzed. Based on the biological characteristics of the fibrous roots being slender, the aspect ratio of the connected region is calculated. If the aspect ratio is greater than a preset morphological threshold, the connected region is determined to be a valid root system region.

[0078] It should be noted that the preset morphology threshold value needs to be set according to the seedling variety and root growth stage. For seedling varieties in the vigorous growth stage with slender fibrous roots, the preset morphology threshold can be set to a higher value, such as 4.0, to strictly eliminate blocky noise and ensure that only typical slender roots are extracted. For varieties in the early growth stage or with shorter and thicker roots, the preset morphology threshold can be appropriately reduced, such as set to 2.0, to prevent missing short roots that have not yet fully elongated. In this embodiment, considering the root morphology under the general root-enhancing container seedling cultivation scenario, the preset morphology threshold is preferably set to 3.0.

[0079] like Figure 4 As shown in the figure, this is a comparative diagram of the root penetration status assessment results. To visually demonstrate the technical effect of this invention, the identification areas are categorized and marked in the figure. The false detection areas of the traditional method represent incorrect identification results resulting from segmentation based solely on initial feature response values. These areas are mainly concentrated in bright interfering objects in the background and inorganic particles at the vents. The effective root areas of the method of this invention represent the effective root areas determined after processing by this method. It can be seen that by introducing a local contrast weight, this invention successfully suppresses background interference from non-container surfaces; by introducing a chromaticity effectiveness coefficient, this invention accurately eliminates perlite interference, which, although bright, exhibits achromatic characteristics, using biological color features. Ultimately, only fibrous roots with biological characteristics and conforming to local contrast logic are retained, achieving accurate monitoring and assessment in complex backgrounds.

[0080] Finally, after screening, the total pixel area of ​​all valid root regions is counted, and the proportion of this total pixel area to the theoretical total area of ​​the vents under the current view is calculated. If this proportion exceeds a preset warning threshold of 5%, it is determined that the root control mechanism of the current root-growth container has failed or that the substrate moisture is too high, and the system automatically generates and outputs a warning signal.

[0081] This concludes the monitoring and evaluation of the root-promoting container seedling status.

[0082] The second aspect of this embodiment provides a root-promoting container seedling status monitoring system, such as... Figure 5As shown, the root-promoting container seedling status monitoring system includes a memory and a processor. The memory stores computer program instructions, and when the computer program instructions are executed by the processor, a root-promoting container seedling status monitoring method of the first aspect of the present invention is implemented.

[0083] The root-proliferating container seedling status monitoring system also includes other components well known to those skilled in the art, such as communication buses and communication interfaces. Their settings and functions are known in the art and will not be described in detail here.

[0084] In this invention, the aforementioned memory can be any tangible medium containing or storing a program that can be used or combined with an instruction execution system, apparatus, or device. For example, a computer-readable storage medium can be any suitable magnetic or magneto-optical storage medium, such as resistive random access memory (DRAM), dynamic random access memory (DRAM), static random access memory (SRAM), enhanced dynamic random access memory (DRAM), high-bandwidth memory, hybrid memory cube, etc., or any other medium that can be used to store desired information and can be accessed by an application, module, or both. Any such computer storage medium can be part of a device or accessible to or connected to a device.

[0085] The above are all preferred embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Therefore, all equivalent changes made in accordance with the structure, shape and principle of the present invention should be covered within the scope of protection of the present invention.

Claims

1. A method of monitoring the state of propagation in a root enhancing container, characterized by, The method comprises the following steps: An image of the side wall of the root-increasing container is collected, and the image of the side wall is converted to a Lab color space to obtain a Lab image; Global color statistical features of the Lab image are calculated; for each pixel point in the Lab image, an initial feature response value of the pixel point is determined based on a color difference measurement of a color vector of the pixel point and the global color statistical features; A local contrast weight of the pixel point is determined based on a difference between a brightness value of the pixel point and a brightness mean value in a neighborhood window of the pixel point, and the brightness mean value of the neighborhood window, and the relationship is satisfied: ； wherein, is a local contrast weight of the th pixel point; is a luminance value of the th pixel point; is a luminance mean value of a neighborhood window in which the th pixel point is located; is a preset luminance reference value; is a preset infinitesimal value; is a base weight coefficient; is a natural exponential function; A chroma of the pixel point is calculated, the local contrast weight is corrected by using the chroma to obtain a chroma effective coefficient of the pixel point, and the relationship is satisfied: ； in, It is the first The effective chromaticity coefficient of each pixel; It is the first Local contrast weights for each pixel; , They are the first The red-green and yellow-blue components of each pixel in the Lab color space; It is a natural exponential function; The initial feature response value and the chroma effective coefficient are multiplied to obtain a final feature response value of the pixel point; and whether the root system penetrates out is determined based on the final feature response value to evaluate the seedling raising state of the root-increasing container.

2. The method of claim 1, wherein the method comprises: The local contrast weight of the pixel point is determined based on the difference between the brightness value of the pixel point and the brightness mean value in the neighborhood window of the pixel point, and the brightness mean value of the neighborhood window, and the relationship is satisfied: A relative contrast coefficient is obtained by calculating a ratio of the difference and the brightness mean value of the neighborhood window; A brightness modulation coefficient is determined based on the brightness mean value of the neighborhood window, wherein the brightness modulation coefficient is negatively correlated with the brightness mean value of the neighborhood window; The local contrast weight is calculated based on a product of the relative contrast coefficient and the brightness modulation coefficient.

3. The method of claim 1, wherein the method comprises: The chroma effective coefficient of the pixel point is obtained by correcting the local contrast weight by using the chroma, and the relationship is satisfied: A chroma adjustment coefficient is determined based on the chroma, wherein the chroma adjustment coefficient is positively correlated with the chroma; The local contrast weight and the chroma adjustment coefficient are multiplied to obtain the chroma effective coefficient.

4. The method of claim 1, wherein the method comprises: The global color statistical features are a global mean vector of the Lab image, and the global color statistical features are obtained by: Extracting a brightness component, a red-green component and a yellow-blue component of each pixel point in the Lab image under the Lab color space; Calculating an arithmetic mean value of the brightness component, the red-green component and the yellow-blue component of all pixel points in the Lab image to obtain a brightness component mean value, a red-green component mean value and a yellow-blue component mean value; The global mean vector is composed of the brightness component mean value, the red-green component mean value and the yellow-blue component mean value.

5. The method of claim 4, wherein the method further comprises: The initial feature response value of the pixel point is determined based on a color difference measurement of a color vector of the pixel point and the global color statistical features, and the relationship is satisfied: The initial feature response value of each pixel point is obtained according to a square of an Euclidean distance between the color vector of each pixel point in the Lab image and the global mean vector.

6. The method of claim 1, wherein the method further comprises: The final feature response value is determined based on the final feature response value to determine whether the root system penetrates out, and the seedling raising state of the root-increasing container is evaluated, and the relationship is satisfied: A gray feature response map is constructed based on the final feature response value of all pixel points, the gray feature response map is binarized and segmented and morphologically processed, and a connected region is extracted; An effective root system region is determined according to geometric morphological features of the connected region, and the seedling raising state of the root-increasing container is evaluated based on an area proportion of the effective root system region.

7. The method of claim 6, wherein the method further comprises: The root growth container seedling raising state is evaluated based on the area proportion of the effective root system area, and the evaluation includes: counting the total area of all pixel points in the Lab image that are determined as the effective root system area; calculating the proportion of the total area to the theoretical total area of the root growth container air holes; in response to the proportion exceeding a preset early warning threshold, generating and outputting an early warning signal.

8. A root augmentation container nursery status monitoring system characterized by, The root growth container seedling raising state monitoring system includes a processor and a memory, and the memory stores computer program instructions that, when executed by the processor, implement a root growth container seedling raising state monitoring method according to any one of claims 1-7.

Images