Image acquisition device for detecting and analyzing growth vigor of rice seedlings in rice seedling bed, operation method of photographing execution mechanism of image acquisition device and method for analyzing growth vigor of rice seedlings in rice seedling bed

Abstract

The invention relates to an image acquisition device for detecting and analyzing the growth vigor of rice seedlings in a rice field, a photographing execution mechanism operation method of the device and a rice seedling growth vigor analysis method, in particular to a rice seedling growth vigor analysis device, an operation method of the device and a rice seedling growth vigor analysis method. In order to solve the problems that a traditional manual rice seedling character inspection detection method is large in workload, rice needs to be contacted in the measurement process, and the natural growth process of the rice is affected, firstly, two linear guide rails and a spur rack are vertically and fixedly installed on an inspection robot frame; the shooting auxiliary part has the function that according to the emergence rate, night water spitting recognition, the purpose of measuring and recognizing plant height, leaf width and leaf ear spacing indexes is to monitor the shape indexes of rice seedlings, and the shape indexes of the rice seedlings are measured and recognized according to the height of the rice seedlings, the leaf width and the leaf ear spacing. Character indexes are recognized, the growth state of the rice seedlings is monitored, and management decisions are optimized. The invention belongs to the technical field of rice seedling field seedling character detection.

Description

Technical Field

[0001] The invention belongs to the technical field of rice seedling trait detection, and specifically relates to an image acquisition device for detecting and analyzing the growth of rice seedlings in a rice seedling field, an operation method of a photographing actuator of the device, and a rice seedling growth analysis method in a rice seedling field. Background Art

[0002] Rice seedling raising can shorten the growth period of rice in the field and increase the number of rice plantings on the land. Rice seedling raising can save water, realize centralized management and save costs during the seedling stage, and provide us with high-quality rice seedlings. The plastic floppy disk seedling raising method is also popular now. This seedling raising method is conducive to our mechanized transplanting and throwing, and at the same time, the seedling management is more unified.

[0003] Small seedlings generally refer to seedlings that are transplanted with soil during the 3-leaf stage, and are mostly cultivated on densely sown, heat-insulated seedling beds. Medium seedlings generally refer to seedlings that are transplanted when they have 3-4.5 leaves, and can also be used for early transplanting and broadcasting. Large seedlings generally refer to seedlings that are transplanted when they have 4-6 leaves, and are more suitable for double-season rice and single-season medium-season rice. Strong seedlings with multiple tillers have more leaves, and should not be transplanted until they have 6-9 leaves. They are mostly used for double-season late rice and late-season medium-season rice.

[0004] Whether it is a large seedling or a small seedling, diseases such as seedling blight may occur during the cultivation process. Moreover, if the seedlings are not raised well, rice blast and dwarf disease are more likely to occur in the later planting.

[0005] Effective monitoring of rice seedlings' leaf age, leaf color, tillering and other traits at various growth stages can effectively reflect their growth status, and is of great significance for timely adjusting the temperature, humidity, light and nutrients of the seedling field.

[0006] Traditional methods of detecting rice seedling traits not only consume a lot of manpower and material costs, but are also cumbersome to operate and have a high error rate, which has always been a major problem restricting rice production regulation. Summary of the invention

[0007] The purpose of the present invention is to solve the problems of large workload of traditional manual inspection seedling trait detection methods and the need to contact rice during the measurement process, which affects the natural growth process of rice. A rice seedling growth analysis device and a seedling trait analysis method are provided.

[0008] An image acquisition device for detecting and analyzing the growth of rice seedlings in a rice field comprises a power supply control unit, a track sliding vehicle, a photographing actuator, a central control unit and an automatic photographing system.

[0009] The power supply control unit supplies power to the rail vehicle, the photographing actuator, the central control unit and the automatic photographing system. The central control unit controls the operation of the rail vehicle, the photographing actuator and the automatic photographing system. The automatic photographing system is installed on the inspection robot frame of the rail vehicle through the photographing actuator.

[0010] Furthermore, the photographing actuator includes a main body plate, a lifting assembly, a telescopic assembly, a seedling-moving assembly, a photographing auxiliary part, two linear guide rails and a main body plate guide rail slider;

[0011] The lifting component is installed on the main body plate, the main body plate is slidably arranged on two linear guide rails through two main body plate guide sliders, the telescopic component is slidably arranged on the main body plate, the seedling-moving component is installed on the telescopic component, and the photographing auxiliary part is installed on the main body plate.

[0012] Further, the lifting assembly includes a spur rack, a lifting servo motor, a lifting reducer and a lifting gear;

[0013] The telescopic assembly includes a slide plate, a telescopic linear guide rail, a telescopic synchronous toothed belt, a telescopic reducer and a telescopic servo motor;

[0014] The seedling pulling component includes a seedling pulling synchronous toothed belt, a seedling pulling linear guide rail, a right pull rod, a left pull rod, a seedling pulling servo motor and a seedling pulling reducer;

[0015] The photographing auxiliary parts include a photographing baffle and a photographing electromagnet;

[0016] Two linear guide rails are vertically fixed on the inspection robot frame, and a spur rack is vertically fixed on the inspection robot frame. The lifting servo motor is connected to the lifting gear through a lifting reducer. The lifting gear is driven to rotate by the lifting servo motor. Two main plate guide sliders are installed on the main plate, and the lifting gear is meshed with the spur rack teeth.

[0017] The telescopic linear guide is installed on the bottom end surface of the main body plate, the guide slider of the slide plate is slidably set on the telescopic linear guide, the telescopic servo motor and the telescopic reducer are installed on the upper end surface of the main body plate, the telescopic servo motor drives the active telescopic synchronous pulley on the telescopic synchronous toothed belt to rotate through the telescopic reducer, and the driven telescopic synchronous pulley is rotatably connected and set on the bottom end surface of the main body plate, the telescopic synchronous toothed belt sleeve is set on the active telescopic synchronous pulley and the driven telescopic synchronous pulley, and the length direction of the telescopic synchronous toothed belt is set in the same direction as the length square of the telescopic linear guide, the slide plate is fixed to the single-side toothed belt of the synchronous toothed belt through the fixed seat on the slide plate, and the slide plate is driven by the synchronous toothed belt to slide along the telescopic linear guide.

[0018] The seedling-pulling linear guide is fixedly mounted on the bottom end surface of the slide plate, and the right lever and the left lever are slidably arranged on the seedling-pulling linear guide rail. The seedling-pulling servo motor and the seedling-pulling seedling pulley rotate with the active seedling-pulling synchronous pulley of the seedling-pulling synchronous toothed belt through the reducer, and the driven seedling-pulling synchronous pulley is rotatably connected and arranged on the bottom end surface of the slide plate. The seedling-pulling synchronous toothed belt is sleeved on the active seedling-pulling synchronous pulley and the driven seedling-pulling synchronous pulley. The fixed seat of the right lever is fixed to the toothed belt on one side of the seedling-pulling synchronous toothed belt, and the fixed seat of the left lever is fixed to the toothed belt on the other side of the seedling-pulling synchronous toothed belt. The right lever and the left lever are driven by the seedling-pulling synchronous toothed belt to slide toward or away from each other along the length direction of the seedling-pulling linear guide rail.

[0019] A spring push rod is provided above the photo electromagnet, and the photo baffle is fixedly connected to the spring push rod above the photo electromagnet. The photo electromagnet is electrified to adsorb the spring push rod to drive the photo baffle to move downward, and the photo electromagnet is de-energized and the spring push rod is away from the photo electromagnet to drive the photo baffle to move upward.

[0020] Furthermore, the automatic photographing system includes a first camera and a second camera. The first camera is installed on the top of the photographing actuator, and the second camera is fixedly installed below the photographing actuator. The central control unit controls the first camera and the second camera to perform photographing.

[0021] A method for operating a photographing actuator of a rice seedling image acquisition device in a rice seedling field, the method being implemented in the following manner:

[0022] First, two linear guide rails and spur racks are vertically fixed on the inspection robot frame;

[0023] Lifting of the camera actuator: The lifting servo motor and the lifting reducer drive the lifting gear to rotate on the spur rack, so that the main plate moves linearly on the linear guide according to the main plate guide slider to realize the lifting or lowering action of the actuator;

[0024] Telescopic operation of the camera actuator: the telescopic servo motor and the telescopic reducer drive the active telescopic synchronous belt wheel on the telescopic synchronous belt to rotate, and the telescopic synchronous belt drives the driven telescopic synchronous belt wheel to force the telescopic synchronous belt wheel to rotate around the active telescopic synchronous belt wheel and reciprocate the driven telescopic synchronous belt wheel, and the telescopic synchronous belt drives the slide plate to slide along the telescopic linear guide rail through the telescopic synchronous belt, so as to realize the reciprocating motion of the slide plate on the telescopic linear guide rail, i.e. the telescopic operation of the actuator;

[0025] The seedlings pulling of the photographing actuator: the active seedlings pulling synchronous pulley is driven to rotate by the seedlings pulling servo motor and the seedlings pulling through the reducer, and the driven seedlings pulling synchronous pulley is driven to rotate by the seedlings pulling synchronous toothed belt, forcing the seedlings pulling synchronous toothed belt to rotate around the active seedlings pulling synchronous pulley and reciprocate the driven seedlings pulling synchronous pulley, and a slider with a right lever and a slider with a left lever slide on the seedlings pulling linear guide rail. When the seedlings pulling servo motor rotates, the toothed belts on both sides of the seedlings pulling synchronous toothed belt are driven to move in the opposite direction, thereby realizing that the seedlings pulling synchronous toothed belt drives the right lever and the left lever to slide toward or away from each other along the length direction of the seedlings pulling linear guide rail;

[0026] The function of the photo-taking auxiliary parts: In order to obtain a good image of the shape of the rice seedlings, the quick response of the electromagnet is used to drive the photo-taking baffle to move back and forth. The photo-taking baffle is driven downward by the charged adsorption spring push rod of the photo-taking electromagnet, and the photo-taking baffle is driven to serve as the background of the image. When not taking pictures, the photo-taking electromagnet is powered off and the spring push rod drives the photo-taking baffle to move upward and lift up.

[0027] A method for analyzing the growth of rice seedlings in a rice field, the method comprising the following steps:

[0028] Step 1: The first camera is installed on the top of the rice seedling growth analysis device. The height and position of the first camera are adjusted so that the first camera is located above the rice seedlings in the rice seedling field. The first camera is used to image the target rice seedlings and calculate the emergence rate.

[0029] Step 2: The second camera is installed at the bottom of the rice seedling growth analysis device. The rice seedling growth analysis device is controlled to adjust the shooting angle of the second camera to achieve timed shooting to obtain a rice image. The collected target rice image is used to perform seedling water spitting detection using a threshold segmentation method. In the process of seedling water spitting detection, the probability is obtained according to the grayscale histogram of the rice image to select the threshold and the number of segmentations, and the water spitting area of ​​the seedling is obtained to complete the water spitting identification;

[0030] Step 3: controlling the motor of the rice seedling growth analysis device, pushing aside the rice seedlings and capturing the rice image through the second camera;

[0031] Based on the collected rice images, the connected domain algorithm is first used to extract individual seedlings, and the individual seedlings are grayed and binarized to obtain a grayscale image. The grayscale image is then closed and filled with small holes and isolated points in the image. The contour is extracted, and the "convex hull" is detected from the contour. A straight line from the "convex defect" to the bottom of the image is drawn and the distance is calculated. The number of leaves is determined based on the number of "convex defects" and the distance from the "convex defect" to the bottom of the image, and leaf age recognition is performed.

[0032] Step 4: controlling the motor of the rice seedling growth analysis device, pushing aside the rice seedlings and capturing the rice image through the second camera;

[0033] Based on the collected rice image, first use the connected domain algorithm to extract individual seedlings, and then grayscale and binarize the individual seedlings to obtain a grayscale image, and then perform a closed operation on the grayscale image to find the "convex defect"; or according to the rice image collected in step 3, perform a closed operation to find the "convex defect" to complete the calculation of the ear spacing:

[0034] The calculation of the distance between the ear of the plant is based on the convex defect. The convex defect points corresponding to the first leaf and the second leaf are found, and the pixel area of ​​the main stem of the seedling connecting the two points is calculated, and then the actual length of the stem and leaves between the two points of the main stem of the seedling is determined as the distance between the ear of the plant.

[0035] Step 5: Control the motor of the rice seedling growth analysis device, push aside the rice seedlings and capture the rice image through the second camera, extract the single rice seedlings based on the captured rice image using the connected domain algorithm, and grayscale and binarize the single rice seedlings to obtain a grayscale image, or process the single rice seedlings according to step 3 to obtain a grayscale image, perform expansion and corrosion operations on the obtained grayscale image, and use the corner detection method to detect the leaf width of the rice seedlings:

[0036] On the grayscale image, a point pair is formed based on each corner point and the corner point closest to it, and the average distance of the point pairs is calculated. Specifically, the distances of adjacent corner point pairs whose pixel distance is less than the pixel threshold condition are stored in a list, and these point pairs are connected with line segments on the image. The average width of the point pairs is calculated based on the effective distance, which is used as the final leaf width.

[0037] Step 6: As the rice seedlings grow, the motor of the rice seedling growth analysis device is controlled, the rice seedlings are pushed aside and photographed by a second camera to obtain a rice image, based on the collected rice image, a single rice seedling is extracted using a connected domain algorithm, and the single rice seedling is grayed and binarized to obtain a grayscale image, and the grayscale image is expanded and eroded, or the grayscale image obtained by the single rice seedling is expanded and eroded according to step 5, and the plant height of the stem and leaves is calculated using the rice image;

[0038] The purpose of measuring and identifying the seedling emergence rate, night water spitting identification, plant height, leaf width, and leaf ear spacing indicators is to monitor the shape indicators of rice seedlings, and to monitor the growth status of rice seedlings and optimize management decisions by identifying these trait indicators.

[0039] Furthermore, the process of detecting water spitting of rice seedlings by using the threshold segmentation method in step 2 is implemented by using the Ostu algorithm:

[0040] First, calculate the grayscale histogram and its normalized histogram of the image, assuming that the grayscale ranges from 0 to L-1, and the number of pixels with grayscale level i is n i, the total number of pixels is N, then the probability of gray level i is P(i):

[0041]

[0042] Then calculate the intra-class variance and inter-class variance, and finally select the threshold t that maximizes the inter-class variance as the segmentation threshold. The intra-class variance and inter-class variance are calculated by the following formulas:

[0043]

[0044] where ω0(t) and ω1(t) are the cumulative distributions of background and foreground pixels:

[0045]

[0046] and are the variances of background and foreground pixels respectively:

[0047]

[0048] μ0(t) and μ1(t) are the average grayscale values ​​of background and foreground pixels:

[0049]

[0050] Based on the segmentation threshold, the image is segmented to determine the water spouting area.

[0051] Furthermore, in step 3, the connected domain algorithm is used to extract individual seedlings and the grayscale formula of the RGB image for grayscale and binarization of the individual seedlings is:

[0052] I gray (x,y)=0.299R(x,y)+0.587G(x,y)+0.114B(x,y)

[0053] Binarization formula:

[0054]

[0055] Where T is the binarization threshold,

[0056] The distance from the "convex defect" to the bottom of the image in step 3 is calculated using the Euclidean distance formula:

[0057]

[0058] Where (x1,y1) is the point of the convex defect and (x2,y2) is the corresponding point at the bottom of the image.

[0059] Furthermore, step five calculates the average width of the leaves according to the effective distance as the final leaf width:

[0060] The steps of calculating leaf width based on corner detection are as follows: first, skeleton extraction of a single seedling is performed to obtain a binary image, then the erosion and dilation operation is used to smooth the edges, and then the corner detection algorithm is used to mark the turning points in the image.

[0061] Let the corner point be P i (x i ,y i ), whose nearest right corner is P j (x j ,y j ), the distance calculation formula can be:

[0062]

[0063] If d ij <20 pixels, store it in the list,

[0064] Assume that the distance list of corner points is {d1,d2,...,d n}, the total distance is The calculation formula for the blade width is:

[0065]

[0066] n is the number of corner point pairs.

[0067] Furthermore, in step 6, changing the second camera to complete the detection of plant height includes the following steps:

[0068] Step 61: First, separate the rice seedlings and leaves;

[0069] Step 62: convert the separated rice seedlings and leaves into binary images;

[0070] Step 63: Extract skeleton from the converted binary image;

[0071] Step 64: Perform convolution expansion operation on the obtained skeleton;

[0072] Step 65: Calculate the proportional factor by the ratio of the pixel area of ​​the expanded skeleton image to the pixel area of ​​the marker, and calculate the true height of the stem and leaf by the proportional factor and the pixel area of ​​the expanded skeleton image.

[0073] Beneficial effects:

[0074] 1. This application adopts a rice seedling image acquisition device with the functions of walking, taking pictures, and adjusting its angle according to the growth of rice seedlings. The active visual trait detection method combined with neural network and deep learning is of great significance.

[0075] 2. The rice seedling image acquisition device of the present application has the ability to control and automatically walk on a predetermined track, take pictures when it runs to a specified position, and the data after taking pictures has the ability to be stored and remotely transmitted. Moreover, it can monitor the growth of rice seedlings in real time, and the rice seedling image acquisition device can adjust up and down and front and back to take pictures. It can reproduce the growth and leaf age of actual rice seedlings. The collected images obtain the ear spacing, leaf age, leaf width, plant height, and night water spitting of rice seedlings, and combine neural networks and artificial intelligence to realize intelligent regulation and analysis and calculation of the final emergence rate. Moreover, the rice seedling trait analysis method of the present invention does not need to contact rice during the measurement process, and will not affect the natural growth process of rice. BRIEF DESCRIPTION OF THE DRAWINGS

[0076] Figure 1 This is a schematic diagram of the composition of the intelligent inspection robot for rice seedlings in rice fields;

[0077] Figure 2 A schematic diagram of the direction of the photo execution mechanism overlooking the sky;

[0078] Figure 3 This is a schematic diagram of the actuator side viewed from above;

[0079] Figure 4 A schematic diagram of the actuator for taking photos from above;

[0080] Figure 5 This is a schematic diagram of the inspection robot framework;

[0081] Figure 6 It is the overall flow chart of the active visual trait rice seedling trait detection method;

[0082] Figure 7 This is the effect diagram of seedling emergence rate detection;

[0083] Figure 8 This is the effect diagram of the seedling water spitting test;

[0084] Fig. 9 Schematic diagram of extracting individual seedlings from the rice field;

[0085] Fig.10 Schematic diagram of finding convex defects for rice seedlings in rice fields;

[0086] Fig.11 This is a schematic diagram of detecting the corner points of rice seedlings in a rice field;

[0087] Fig.12 This is a schematic diagram of detecting the height of seedlings in the rice field. DETAILED DESCRIPTION Specific implementation method one:

[0089] Combination Figure 1-Figure 5This embodiment is an image acquisition device for detecting the growth analysis of rice seedlings in rice fields. It includes a power supply control unit, a track slide, a photo actuator, a central control unit and an automatic photo system.

[0090] The power supply control unit supplies power to the rail trolley, the photographing actuator, the central control unit and the automatic photographing system. The central control unit controls the operation of the rail trolley, the photographing actuator and the automatic photographing system. The automatic photographing system is installed on the rail trolley through the photographing actuator.

[0091] The power supply control unit supplies power to the entire analysis device and has the functions of remote (user end) direct control of timed on / off and automatic charging;

[0092] The track trolley can be controlled by a mobile phone APP to move forward, backward, slow down and fast on the track, and has a power-off function in case of faults, and completes automatic inspection in the greenhouse through the central control unit;

[0093] The camera actuator can complete timed photography through remote (user end) and on-site control of the image acquisition camera. It has the functions of lifting, extending, retracting and pushing the seedlings, and can adjust the distance in all directions in real time according to the growth of the seedlings. The camera actuator contains three motors, namely lifting, extending, retracting and pushing the seedlings, and the three motors have independent degrees of freedom. The lifting direction has an anti-violence function. It should be noted that the downward probe is tested on flat ground. It is easy to be damaged if it dives too deep. Therefore, the probe of the camera actuator needs to be tested at a height of 10cm from the ground.

[0094] The photographing actuator includes a main plate 6, a lifting assembly, a telescopic assembly, a seedling-moving assembly, a photographing auxiliary part, two linear guide rails 1 and a main plate guide rail slider 2;

[0095] The lifting assembly is installed on the main plate 6, and the main plate 6 is slidably set on two linear guide rails 1 through two main plate guide sliders 2. The telescopic assembly is slidably installed on the main plate 6, the seedling moving assembly is installed on the telescopic assembly, and the photo-taking auxiliary part is installed on the main plate 6.

[0096] The lifting assembly includes a spur rack 3, a lifting servo motor 4, a lifting reducer 5 and a lifting gear 17;

[0097] The telescopic assembly includes a slide plate 8, a telescopic linear guide rail 12, a telescopic synchronous toothed belt 13, a telescopic reducer 15 and a telescopic servo motor 16;

[0098] The seedling-pushing assembly includes a seedling-pushing synchronous toothed belt 7, a seedling-pushing linear guide rail 9, a right lever 10, a left lever 11, a seedling-pushing servo motor 18 and a seedling-pushing reducer 19;

[0099] The photographing auxiliary parts include a photographing baffle 20 and a photographing electromagnet 21;

[0100] Two linear guide rails 1 are vertically fixedly installed on the rail trolley, and a spur rack 3 is vertically fixedly installed on the rail trolley. A lifting servo motor 4 is connected to a lifting gear 17 through a lifting reducer 5. The lifting gear 17 is driven to rotate by the lifting servo motor 4. Two main body plate guide sliders 2 are installed on the main body plate 6. The lifting gear 17 is meshed with the spur rack 3.

[0101] The telescopic linear guide 12 is installed on the bottom end surface of the main plate 6, the guide slider of the slide plate 8 is slidably set on the telescopic linear guide 12, the telescopic servo motor 16 and the telescopic reducer 15 are installed on the upper end surface of the main plate 6, the telescopic servo motor 16 drives the active telescopic synchronous pulley on the telescopic synchronous toothed belt 13 to rotate through the telescopic reducer 15, and the driven telescopic synchronous pulley is rotatably connected and set on the bottom end surface of the main plate 6, the telescopic synchronous toothed belt 13 is sleeved on the active telescopic synchronous pulley and the driven telescopic synchronous pulley, and the length direction of the telescopic synchronous toothed belt 13 is set in the same direction as the length square of the telescopic linear guide 12, the slide plate 8 is fixed to the single-sided toothed belt of the synchronous toothed belt 13 through the fixed seat on the slide plate 8, and the slide plate 8 is driven by the synchronous toothed belt 13 to slide along the telescopic linear guide 12.

[0102] The seedling-pulling linear guide 9 is fixedly mounted on the bottom end surface of the slide plate 8, and the right lever 10 and the left lever 11 are slidably arranged on the seedling-pulling linear guide 9. The seedling-pulling servo motor 18 and the seedling-pulling seedling are rotated with the active seedling-pulling synchronous pulley of the seedling-pulling synchronous toothed belt 7 through the reducer 19, and the driven seedling-pulling synchronous pulley is rotatably connected and arranged on the bottom end surface of the slide plate 8. The seedling-pulling synchronous toothed belt 7 is sleeved on the active seedling-pulling synchronous pulley and the driven seedling-pulling synchronous pulley. The fixed seat of the right lever 10 is fixed to the toothed belt on one side of the seedling-pulling synchronous toothed belt 7, and the fixed seat of the left lever 11 is fixed to the toothed belt on the other side of the seedling-pulling synchronous toothed belt 7. The right lever 10 and the left lever 11 are driven by the seedling-pulling synchronous toothed belt 7 to slide toward or away from each other along the length direction of the seedling-pulling linear guide 9.

[0103] A spring push rod is provided above the photo electromagnet 21, and the photo baffle 20 is fixedly connected to the spring push rod above the photo electromagnet 21. The photo electromagnet 21 is electrically attracted to the spring push rod to drive the photo baffle 20 to move downward. When the photo electromagnet 21 is powered off, the spring push rod moves away from the photo electromagnet 21 to drive the photo baffle 20 to move upward.

[0104] The automatic photography system includes a first camera and a second camera 14. The first camera is installed at the top of the photography actuator, and the second camera 14 is fixedly installed below the photography actuator. The central control unit controls the first camera and the second camera 14 to perform photography. The first camera is a ball camera, and the second camera 14 is an image acquisition camera.

[0105] The lifting stroke of the lifting component of the photographing actuator is 150mm, the telescopic component stroke is 150mm, and the seedling moving component stroke is 150mm.

[0106] The rice seedling detection process of the photographing actuator is descending, extending, descending, extending, the lever opens, the baffle falls, photographing, the baffle rises, the lever closes, rises, and retracts until the cycle ends.

[0107] The central control unit serves as the control center of the entire rice seedling field growth analysis device. It has the functions of automatic software update, remote control of the track slide car to complete automatic inspection in the greenhouse, remote control of timed photography, real-time transmission of image data after photography (that is, automatic synchronization of data for the rice seedling field controller), remote control of the servo system of the photography actuator to set the horizontal and vertical offset (zero point calibration, screen setting, mechanical setting), remote setting of preset photography stop points, inspection time (positioning function) and real-time display of collected images and analysis results.

[0108] The first camera of the automatic photography system completes the top photography, receives the command from the central control unit, stretches a certain distance, and realizes the vertical seedling bed photography; the second camera 14 completes the bottom photography, realizes the bottom side photography and has a wide-angle function, and is equipped with a ruler to complete the measurement of the average plant height, etc. Such a configuration can complete both single plant observation and group monitoring, and enables the device to complete the detection of single plant traits such as leaf age, plant height, leaf ear spacing, leaf width, and group detection of leaf age range. Specific implementation method 2:

[0110] Combination Figure 2-Figure 4 This embodiment is a method for operating a photographing actuator of a rice seedling image acquisition device in a rice seedling field, and the method is implemented according to the following steps:

[0111] First, two linear guide rails 1 and spur rack 3 are vertically fixedly mounted on the inspection robot frame as a prerequisite for the working of the actuator;

[0112] Lifting of the photographing actuator: The lifting servo motor 4 and the lifting reducer 5 drive the lifting gear 17 to rotate on the spur rack 3, so that the main plate 6 moves linearly on the linear guide 1 according to the main plate guide slider 2, so as to realize the lifting or lowering action of the actuator;

[0113] Telescopic operation of the photographing actuator: the telescopic servo motor 16 and the telescopic reducer 15 drive the active telescopic synchronous belt wheel on the telescopic synchronous toothed belt 13 to rotate, and the telescopic synchronous toothed belt 13 drives the driven telescopic synchronous belt wheel to force the telescopic synchronous belt 13 to rotate around the active telescopic synchronous belt wheel and reciprocate the driven telescopic synchronous belt wheel, and the telescopic synchronous toothed belt 13 drives the slide plate 8 to slide along the telescopic linear guide rail 12, so as to realize the reciprocating motion of the slide plate 8 on the telescopic linear guide rail 12, that is, the telescopic function of the actuator;

[0114] The seedlings-pulling mechanism of the photographing actuator: the active seedlings-pulling synchronous pulley is driven to rotate by the seedlings-pulling servo motor 18 and the seedlings-pulling reducer 19, and the driven seedlings-pulling synchronous pulley is driven to rotate by the seedlings-pulling synchronous toothed belt 7, forcing the seedlings-pulling synchronous toothed belt 7 to rotate around the active seedlings-pulling synchronous pulley and reciprocate the driven seedlings-pulling synchronous pulley, and a slider with a right lever 10 and a slider with a left lever 11 slide on the seedlings-pulling linear guide 9. When the seedlings-pulling servo motor 18 rotates, the toothed belts on both sides of the seedlings-pulling synchronous toothed belt 7 are driven to move in the opposite direction, thereby realizing that the seedlings-pulling synchronous toothed belt 7 drives the right lever 10 and the left lever 11 to slide toward or away from each other along the length direction of the seedlings-pulling linear guide 9.

[0115] Function of the photographing auxiliary parts: In order to obtain a good image of the shape of the rice seedlings, the quick response of the electromagnet is utilized to drive the photographing baffle 20 to move back and forth, and the photographing baffle 20 is driven downward by the charged adsorption spring push rod of the photographing electromagnet 21, and driven to serve as the background of the image. When not taking pictures, the photographing electromagnet 21 is powered off and the spring push rod drives the photographing baffle 20 to move upward and lift up. Specific implementation method three:

[0117] Combination Figure 6-Figure 12 The method for analyzing the growth of rice seedlings in a rice field according to the present embodiment is described, and the method comprises the following steps:

[0118] Step 1: The first camera is installed on the top of the rice seedling growth analysis device. The height and position of the first camera are adjusted so that the first camera is located above the rice seedlings in the rice seedling field. The first camera is used to image the target rice seedlings and calculate the emergence rate.

[0119] Read the image and select a rectangular area, convert the color space of the rectangular area from RGB to HSV to separate the green pixels, calculate the proportion of green (green pixels divided by all pixels in the rectangular area) to get the seedling rate, and use the image collected in step 1 as shown in Figure 7 As shown, the formula for calculating the emergence rate is:

[0120]

[0121] Step 2: The second camera 14 is installed at the bottom of the rice seedling growth analysis device. The rice seedling growth analysis device is controlled to adjust the shooting angle of the second camera 14 to achieve timed shooting to obtain rice images. The second camera 14 is controlled to shoot according to the pre-set time control device. The shooting position is adjusted by controlling the motor to move forward and backward and up and down. Figure 8 As shown, the rice image obtained by taking a photo with the second camera 14 can be used to identify the rice seedlings spitting water during the day and at night:

[0122] The target rice image collected is used to detect water spitting in rice seedlings using the threshold segmentation method. In the process of water spitting detection in rice seedlings, the probability is obtained according to the grayscale histogram of the rice image to select the threshold and the number of segmentations, and the water spitting area of ​​the rice seedlings is obtained to complete the water spitting identification:

[0123] The process of detecting seedling water spitting using the threshold segmentation method is implemented using the Ostu algorithm:

[0124] First, calculate the grayscale histogram and its normalized histogram of the image, assuming that the grayscale ranges from 0 to L-1, and the number of pixels with grayscale level i is n i , the total number of pixels is N, then the probability of gray level i is P(i):

[0125]

[0126] Then calculate the intra-class variance and inter-class variance, and finally select the threshold t that maximizes the inter-class variance as the segmentation threshold. The intra-class variance and inter-class variance can be calculated by the following formulas:

[0127]

[0128] where ω0(t) and ω1(t) are the cumulative distributions of background and foreground pixels:

[0129]

[0130] and are the variances of background and foreground pixels respectively:

[0131]

[0132] μ0(t) and μ1(t) are the average grayscale values ​​of background and foreground pixels:

[0133]

[0134] The threshold segmentation algorithm uses the obtained threshold to divide the grayscale of the seedling image into 2 or 3 parts, which are background, seedlings and water drops during the day; at night, the background and seedlings are regarded as the same part, so there are 2 parts at night, and the pixels belonging to the same part are the same object.

[0135] Step 3: Control the motor of the rice seedling growth analysis device, push aside the rice seedlings and capture the rice image through the second camera 14;

[0136] Based on the collected rice images, the connected domain algorithm is first used to extract individual rice seedlings. Based on the collected rice images, the connected domain algorithm is first used (one closed domain is considered a connected domain, different connected domains are distinguished by different colors, and the connected domain with the smallest area is selected as the target) to extract individual rice seedlings, and the individual rice seedlings are grayed and binarized to obtain a grayscale image, and then the grayscale image is closed. Fig. 9 and Fig.10 Shown

[0137] Grayscale formula for RGB image:

[0138] I gray (x,y)=0.299R(x,y)+0.587G(x,y)+0.114B(x,y)

[0139] Binarization formula:

[0140]

[0141] Where T is the binarization threshold,

[0142] Then, a closing operation is performed to fill the small holes and isolated points in the image to make the target area more connected, extract the contour, and then detect the "convex hull" from the contour (the convex hull is a convex polygon formed by connecting the outermost points). The "concave" area, that is, the "convex defect" is found from the convex hull (OpenCV provides the function convexHull for detecting the convex hull of the object contour, and the function convexityDefect for detecting convex defects). Draw a straight line from the "convex defect" to the bottom of the image and calculate the distance. A convex hull will be formed between a leaf and the trunk, but the trunk is not necessarily smooth and may have small protrusions, which will also be considered as a convex hull, so the distance to the bottom must be added. The distance between leaves is roughly in a range. If the distances of several convex hulls to the bottom are very close, they can be regarded as a convex hull. The number of leaves is determined according to the number of "convex defects" and the distance from the "convex defect" to the bottom of the image. If the number of leaves is 1, the seedling is in the one-leaf stage.

[0143] The number of leaves is determined based on the number of "convex defects" and the distance from the "convex defects" to the bottom of the image. The number of "convex defects" that meet the set distance is the number of leaves:

[0144] The distance from the "convex defect" to the bottom of the image can be calculated using the Euclidean distance formula:

[0145]

[0146] Where (x1,y1) is the point of the convex defect and (x2,y2) is the corresponding point at the bottom of the image.

[0147] like Fig.10 and Fig.11 As shown, step 4: controlling the motor of the rice seedling growth analysis device, pushing aside the rice seedlings and capturing the rice image through the second camera 14; based on the captured rice image, firstly using the connected domain algorithm to extract the single rice seedlings, and graying and binarizing the single rice seedlings to obtain a grayscale image, and then performing a closed operation on the grayscale image to find the "convex defect"; or performing a closed operation on the rice image obtained in step 3 to find the "convex defect" to complete the calculation of the ear spacing:

[0148] The calculation of the distance between the ear of the rice seedling is based on the convex defect, and the convex defect points corresponding to the first leaf and the second leaf are found, wherein the first leaf refers to the first leaf closest to the root, which is usually the earliest leaf to grow. The second leaf refers to the second leaf from the root, which is usually the second earliest leaf to grow. The pixel area of ​​the main stem of the rice seedling connecting the two points is calculated and compared with the pixel area of ​​the measuring ruler, and the actual length of the stem and leaves between the two points of the main stem of the rice seedling is calculated as the distance between the ear of the rice seedling; the measuring ruler is a plane or a plate with a scale, and the measuring ruler is set on the camera baffle 20.

[0149] Step 5: Control the motor of the rice seedling growth analysis device, push the rice seedlings aside and capture the rice image through the second camera 14, extract the single rice seedlings based on the captured rice image using the connected domain algorithm, and grayscale and binarize the single rice seedlings to obtain a grayscale image, or process the single rice seedlings according to step 3 to obtain a grayscale image, perform expansion and corrosion operations on the obtained grayscale image, and use the corner detection method to detect the leaf width of the rice seedlings:

[0150] On the grayscale image, a point pair is formed based on each corner point and the corner point closest to it, and the average distance of the point pairs is calculated. Specifically, the distances of adjacent corner point pairs whose pixel distance is less than the pixel threshold condition are stored in a list, and these point pairs are connected with line segments on the image. The average width of the point pairs is calculated based on the effective distance, which is used as the final leaf width.

[0151] Corner points are confirmed by detecting small areas with significant grayscale changes and variable directions in the image based on the grayscale image obtained in step 3. This method finds the nearest adjacent corner point to each corner point and calculates the average distance. Specifically, the distances of adjacent corner point pairs with a distance less than 20 pixels are stored in a list, and these point pairs are connected with blue lines on the original image. The average width of the leaf is calculated based on the effective distance. Here, the total distance is divided by 3 times the number of corner point pairs to obtain the final leaf width.

[0152] Let the corner point be P i (x i ,y i ), whose nearest right corner is P j (x j ,y j ), the distance calculation formula can be:

[0153]

[0154] If d ij <20 pixels, store it in a list.

[0155] Assume that the distance list of corner points is {d1,d2,...,d n}, the total distance is The calculation formula for the blade width is:

[0156]

[0157] Where n refers to the number of corner point pairs, 3n is an empirical value or normalization coefficient used to adjust the final calculated width, and corner detection is the key step in calculating leaf width: first, the skeleton of a single seedling is extracted to obtain a binary image, then the edges are smoothed using dilation and erosion operations, and finally the corner detection algorithm is used to mark the inflection points in the image.

[0158] Step 6: As the rice seedlings grow, the motor of the rice seedling growth analysis device is controlled, the rice seedlings are pushed aside and photographed by the second camera 14 to obtain a rice image. Based on the collected rice image, a single seedling is extracted using a connected domain algorithm, and the single seedling is grayed and binarized to obtain a grayscale image, or a single seedling is processed according to step 3 to obtain a grayscale image, and the obtained grayscale image is expanded and eroded, or a grayscale image obtained by a single seedling is expanded and eroded according to step 5, and the real height of the stems and leaves is calculated using the rice image, such as Fig.12 As shown;

[0159] The purpose of measuring and identifying the seedling emergence rate, night water spitting identification, plant height, leaf width, and leaf ear spacing indicators is to monitor the shape indicators of rice seedlings, which are part of leaf age diagnosis. By identifying these trait indicators, the growth status of rice seedlings can be monitored and management decisions can be optimized.

[0160] In step 6, changing the second camera 14 to complete the detection of plant height includes the following steps:

[0161] Step 61: First, separate the rice seedlings and leaves;

[0162] Step 62: convert the separated rice seedlings and leaves into binary images;

[0163] Step 63: Extract skeleton from the converted binary image;

[0164] Step 64: Perform a convolution expansion operation on the obtained skeleton and specify its expansion coefficient to eliminate the influence of the width of the stems and leaves on the pixel area;

[0165] Step 65: Calculate the proportional factor by the ratio of the pixel area of ​​the expanded skeleton image to the pixel area of ​​the marker, and calculate the true height of the stem and leaf by the proportional factor and the pixel area of ​​the expanded skeleton image.

[0166] Assume that the pixel area of ​​the expanded skeleton image is A 膨胀 , the known pixel area A of the marker in the figure 像素 and the actual area A 实际 , for contrast conversion, the scale factor is:

[0167]

[0168] Finally, calculate the true height of the stems and leaves:

[0169] H 真实 =A 膨胀 × Scale Factor

[0170] The above calculation examples of the present invention are only used to explain the calculation model and calculation process of the present invention in detail, and are not intended to limit the implementation methods of the present invention. For ordinary technicians in the relevant field, other different forms of changes or modifications can be made based on the above description. It is impossible to list all the implementation methods here. All obvious changes or modifications derived from the technical solution of the present invention are still within the scope of protection of the present invention.

Claims

1. An image acquisition device for detecting and analyzing the growth of rice seedlings in a rice field, characterized in that: It includes a power supply control unit, a rail vehicle, a photographing actuator, a central control unit and an automatic photographing system. The power supply control unit supplies power to the rail vehicle, the photographing actuator, the central control unit and the automatic photographing system. The central control unit controls the operation of the rail vehicle, the photographing actuator and the automatic photographing system. The automatic photographing system is installed on the inspection robot frame of the rail vehicle through the photographing actuator.

2. The image acquisition device for detecting the growth analysis of rice seedlings in rice fields according to claim 1, characterized in that: The photographing actuator comprises a main body plate (6), a lifting component, a telescopic component, a seedling pulling component, a photographing auxiliary component, two linear guide rails (1) and a main body plate guide rail slider (2); The lifting assembly is mounted on the main body plate (6), the main body plate (6) is slidably mounted on two linear guide rails (1) through two main body plate guide sliders (2), the telescopic assembly is slidably mounted on the main body plate (6), the seedling-pulling assembly is mounted on the telescopic assembly, and the photographing auxiliary component is mounted on the main body plate (6).

3. The image acquisition device for detecting the growth analysis of rice seedlings in rice fields according to claim 2, characterized in that: The lifting assembly comprises a spur rack (3), a lifting servo motor (4), a lifting reducer (5) and a lifting gear (17); The telescopic assembly comprises a slide plate (8), a telescopic linear guide rail (12), a telescopic synchronous toothed belt (13), a telescopic reducer (15) and a telescopic servo motor (16); The seedling pushing component comprises a seedling pushing synchronous toothed belt (7), a seedling pushing linear guide rail (9), a right pushing rod (10), a left pushing rod (11), a seedling pushing servo motor (18) and a seedling pushing reducer (19); The photographing auxiliary component comprises a photographing baffle (20) and a photographing electromagnet (21); Two linear guide rails (1) are vertically fixedly installed on the inspection robot frame, a spur rack (3) is vertically fixedly installed on the inspection robot frame, a lifting servo motor (4) is connected to a lifting gear (17) through a lifting reducer (5), and the lifting gear (17) is driven to rotate by the lifting servo motor (4), two main body plate guide sliders (2) are installed on the main body plate (6), and the lifting gear (17) is meshed with the spur rack (3). The telescopic linear guide rail (12) is mounted on the bottom end surface of the main body plate (6), the guide rail slider of the slide plate (8) is slidably arranged on the telescopic linear guide rail (12), the telescopic servo motor (16) and the telescopic reducer (15) are mounted on the upper end surface of the main body plate (6), the telescopic servo motor (16) drives the active telescopic synchronous belt wheel on the telescopic synchronous toothed belt (13) to rotate through the telescopic reducer (15), the driven telescopic synchronous belt wheel is rotatably connected and arranged on the bottom end surface of the main body plate (6), the telescopic synchronous toothed belt (13) is sleeved on the active telescopic synchronous belt wheel and the driven telescopic synchronous belt wheel, and the length direction of the telescopic synchronous toothed belt (13) is arranged in the same direction as the length square of the telescopic linear guide rail (12), the slide plate (8) is fixed to the single-side toothed belt of the synchronous toothed belt (13) through a fixed seat on the slide plate (8), and the slide plate (8) is driven by the synchronous toothed belt (13) to slide along the telescopic linear guide rail (12). The seedling-pulling linear guide rail (9) is fixedly mounted on the bottom end surface of the slide plate (8); the right lever (10) and the left lever (11) are slidably mounted on the seedling-pulling linear guide rail (9); the seedling-pulling servo motor (18) and the seedling-pulling device rotate with the active seedling-pulling synchronous pulley of the seedling-pulling synchronous toothed belt (7) through the reducer (19); the driven seedling-pulling synchronous pulley is rotatably connected and mounted on the bottom end surface of the slide plate (8); the seedling-pulling synchronous toothed belt (7) is sleeved on the active seedling-pulling synchronous pulley and the driven seedling-pulling synchronous pulley; the fixed seat of the right lever (10) is fixed to the toothed belt on one side of the seedling-pulling synchronous toothed belt (7); the fixed seat of the left lever (11) is fixed to the toothed belt on the other side of the seedling-pulling synchronous toothed belt (7); the seedling-pulling synchronous toothed belt (7) drives the right lever (10) and the left lever (11) to slide toward or away from each other along the length direction of the seedling-pulling linear guide rail (9); A spring push rod is arranged above the photographing electromagnet (21), and the photographing baffle (20) is fixedly connected to the spring push rod above the photographing electromagnet (21). When the photographing electromagnet (21) is electrified, the spring push rod is attracted to drive the photographing baffle (20) to move downward, and when the photographing electromagnet (21) is de-energized, the spring push rod is away from the photographing electromagnet (21) to drive the photographing baffle (20) to move upward.

4. The image acquisition device for detecting the growth analysis of rice seedlings in rice fields according to claim 1, characterized in that: The automatic photographing system comprises a first camera and a second camera (14), wherein the first camera is mounted on the top of a photographing actuator, and the second camera (14) is fixedly mounted below the photographing actuator, and a central control unit controls the first camera and the second camera (14) to perform photographing operations.

5. A method for operating a photographing actuator of a rice seedling image acquisition device in a rice field, characterized in that: The method is implemented in the following manner: First, two linear guide rails (1) and a spur rack (3) are vertically fixedly mounted on the inspection robot frame; Lifting of the photographing actuator: the lifting servo motor (4) and the lifting reducer (5) drive the lifting gear (17) to rotate on the spur rack (3), so that the main plate (6) moves linearly on the linear guide rail (1) according to the main plate guide slider (2), thereby realizing the lifting or lowering action of the actuator; Telescopic operation of the photographing actuator: the active telescopic synchronous belt wheel on the telescopic synchronous toothed belt (13) is driven to rotate by the telescopic servo motor (16) and the telescopic reducer (15), and the driven telescopic synchronous belt wheel is driven by the telescopic synchronous toothed belt (13) to force the telescopic synchronous toothed belt (13) to rotate around the active telescopic synchronous belt wheel and reciprocate the driven telescopic synchronous belt wheel, and the slide plate (8) is driven to slide along the telescopic linear guide rail (12) by the telescopic synchronous toothed belt (13), so as to realize the reciprocating motion of the slide plate (8) on the telescopic linear guide rail (12), i.e., the telescopic operation of the actuator; The seedling-pulling mechanism of the photographing actuator drives the active seedling-pulling synchronous belt wheel to rotate through the seedling-pulling servo motor (18) and the seedling-pulling speed reducer (19), and drives the driven seedling-pulling synchronous belt wheel to rotate through the seedling-pulling synchronous toothed belt (7), forcing the seedling-pulling synchronous toothed belt (7) to rotate around the active seedling-pulling synchronous belt wheel and reciprocate the driven seedling-pulling synchronous belt wheel, and slidingly providing a slider of a right lever (10) and a slider of a left lever (11) on the seedling-pulling linear guide rail (9), and when the seedling-pulling servo motor (18) rotates, the toothed belts on both sides of the seedling-pulling synchronous toothed belt (7) are driven to move in opposite directions, thereby realizing that the seedling-pulling synchronous toothed belt (7) drives the right lever (10) and the left lever (11) to slide toward or away from each other along the length direction of the seedling-pulling linear guide rail (9); Function of the photographing auxiliary part: in order to obtain a good image of the shape of the rice seedlings, the quick response of the electromagnet is utilized to drive the photographing baffle (20) to move back and forth, and the photographing baffle (20) is driven downward by the charged adsorption spring push rod of the photographing electromagnet (21), so as to drive the photographing baffle (20) as the background of the image. When not taking pictures, the photographing electromagnet (21) is de-energized and the spring push rod drives the photographing baffle (20) to move upward and lift up.

6. A method for analyzing the growth of rice seedlings in a rice field, characterized in that: The method comprises the following steps: Step 1: The first camera is installed on the top of the rice seedling growth analysis device. The height and position of the first camera are adjusted so that the first camera is located above the rice seedlings in the rice seedling field. The first camera is used to image the target rice seedlings and calculate the emergence rate. Step 2: The second camera (14) is installed at the bottom of the rice seedling growth analysis device, and the rice seedling growth analysis device is controlled to adjust the shooting angle of the second camera (14) to achieve timed shooting to obtain a rice image, and the collected target rice image is used to perform seedling water spitting detection by using a threshold segmentation method. During the seedling water spitting detection, a probability is obtained according to a grayscale histogram of the rice image to select a threshold value and a number of segmentations, and the seedling water spitting area is obtained to complete water spitting identification; Step 3: controlling the motor of the rice seedling growth analysis device, pushing aside the rice seedlings and capturing the rice image through the second camera (14); Based on the collected rice images, the connected domain algorithm is first used to extract individual seedlings, and the individual seedlings are grayed and binarized to obtain a grayscale image. The grayscale image is then closed and filled with small holes and isolated points in the image. The contour is extracted, and the "convex hull" is detected from the contour. A straight line from the "convex defect" to the bottom of the image is drawn and the distance is calculated. The number of leaves is determined based on the number of "convex defects" and the distance from the "convex defect" to the bottom of the image, and leaf age recognition is performed. Step 4: controlling the motor of the rice seedling growth analysis device, pushing aside the rice seedlings and capturing the rice image through the second camera (14); Based on the collected rice image, first use the connected domain algorithm to extract individual seedlings, and then grayscale and binarize the individual seedlings to obtain a grayscale image, and then perform a closed operation on the grayscale image to find the "convex defect"; or according to the rice image collected in step 3, perform a closed operation to find the "convex defect" to complete the calculation of the ear spacing: The calculation of the distance between the ear of the plant is based on the convex defect. The convex defect points corresponding to the first leaf and the second leaf are found, and the pixel area of ​​the main stem of the seedling connecting the two points is calculated, and then the actual length of the stem and leaves between the two points of the main stem of the seedling is determined as the distance between the ear of the plant. Step 5: controlling the motor of the rice seedling growth analysis device, pushing aside the rice seedlings and capturing the rice image through the second camera (14); Based on the collected rice images, the connected domain algorithm is used to extract individual rice seedlings, and the individual rice seedlings are grayed and binarized to obtain a grayscale image, or the individual rice seedlings are processed according to step 3 to obtain a grayscale image, and the grayscale image is expanded and eroded, and the corner detection method is used to detect the leaf width of the rice seedlings: On the obtained grayscale image, a point pair is formed according to each corner point and the corner point closest to it, and the average distance of the point pairs is calculated. Specifically, the distances of adjacent corner point pairs whose pixel distance is less than the pixel threshold condition are stored in a list, and these point pairs are connected with line segments on the image. The average width of the point pairs is calculated according to the effective distance as the final leaf width; Step 6: As the rice seedlings in the rice seedling field grow, the motor of the rice seedling growth analysis device is controlled to push aside the rice seedlings and capture the rice image through the second camera (14); Based on the collected rice image, a connected domain algorithm is used to extract a single rice seedling, and the single rice seedling is grayed and binarized to obtain a grayscale image, and the obtained grayscale image is expanded and eroded, or the grayscale image obtained by the single rice seedling is expanded and eroded according to step five, and the plant height of the stem and leaves is calculated using the rice image; The shape indicators of rice seedlings are monitored based on the measurement and identification results of seedling emergence rate, night water spitting identification, plant height, leaf width, and leaf ear spacing. By identifying these trait indicators, the growth status of rice seedlings can be monitored and management decisions can be optimized.

7. The method for analyzing the growth of rice seedlings in a rice field according to claim 6, characterized in that: The process of using threshold segmentation method to detect seedling water spitting in step 2 is implemented by Ostu algorithm: First, calculate the grayscale histogram and its normalized histogram of the image, assuming that the grayscale ranges from 0 to L-1, and the number of pixels with grayscale level i is n i , the total number of pixels is N, then the probability of gray level i is P(i): Then calculate the intra-class variance and inter-class variance, and finally select the threshold t that maximizes the inter-class variance as the segmentation threshold. The intra-class variance and inter-class variance are calculated by the following formulas: where ω0(t) and ω1(t) are the cumulative distributions of background and foreground pixels: and are the variances of background and foreground pixels respectively: μ0(t) and μ1(t) are the average grayscale values ​​of background and foreground pixels: Based on the segmentation threshold, the image is segmented to determine the water spouting area.

8. The method for analyzing the growth of rice seedlings in a rice field according to claim 6, characterized in that: In step 3, the connected domain algorithm is used to extract individual seedlings and grayscale and binarize the RGB image. The grayscale formula is: I gray (x,y)=0.299R(x,y)+0.587G(x,y)+0.114B(x,y) Binarization formula: Where T is the binarization threshold, The distance from the "convex defect" to the bottom of the image in step 3 is calculated using the Euclidean distance formula: Where (x1,y1) is the point of the convex defect and (x2,y2) is the corresponding point at the bottom of the image.

9. The method for analyzing the growth of rice seedlings in a rice field according to claim 6, characterized in that: Step 5: Calculate the average width of the leaves according to the effective distance as the final leaf width: The steps of calculating leaf width based on corner detection are as follows: first, skeleton extraction of individual seedlings is performed to obtain a binary image, then the erosion and dilation operation is used to smooth the edges, and then the corner detection algorithm is used to mark the turning points in the image. Let the corner point be P i (x i ,y i ), whose nearest right corner is P j (x j ,y j ), the distance calculation formula can be: If d ij <20 pixels, store it in the list, Assume that the distance list of corner points is {d1,d2,...,d n }, the total distance is The calculation formula for the blade width is: n is the number of corner point pairs.

10. The method for analyzing the growth of rice seedlings in a rice field according to claim 6, characterized in that: In step 6, changing the second camera (14) to complete the detection of plant height includes the following steps: Step 61: First, separate the rice seedlings and leaves; Step 62: convert the separated rice seedlings and leaves into binary images; Step 63: Extract skeleton from the converted binary image; Step 64: Perform convolution expansion operation on the obtained skeleton; Step 65: Calculate the proportional factor by the ratio of the pixel area of ​​the expanded skeleton image to the pixel area of ​​the marker, and calculate the true height of the stem and leaves by the proportional factor and the pixel area of ​​the expanded skeleton image.

Images