Rice ear chlorophyll estimation method, system, equipment and medium

Abstract

The invention relates to a rice ear chlorophyll estimation method, system and equipment and a medium. The method comprises the following steps: performing image acquisition on a rice ear sample; the method comprises the following steps: preprocessing collected image information of a rice ear sample, inputting a pre-established chlorophyll regression analysis model to predict chlorophyll of the rice ear sample, and estimating the maturity of the rice ear sample based on a chlorophyll prediction result. The rice ear chlorophyll regression analysis model based on the RGB image is adopted, and real-time and nondestructive monitoring of rice ear chlorophyll is achieved. The method can be widely applied to the technical field of crop monitoring.

Description

Technical Field

[0001] The invention belongs to the technical field of crop monitoring, and in particular relates to a method, system, equipment and medium for estimating chlorophyll of rice ears. Background Art

[0002] Rice is one of the main food crops in my country. During the growth and harvesting process of rice, it is very important to determine the correct harvesting time. Timely harvesting can not only ensure high yield, but also maintain the quality of rice ears. Harvesting too early or too late may lead to loss of yield and quality.

[0003] The existing method for detecting chlorophyll content in the panicle is time-consuming and laborious, mainly relying on chemical experiments to obtain data, which is time-consuming and labor-intensive, and highly destructive. It is difficult to meet the requirements of modern agriculture for efficiency, intelligence, and precision. Therefore, the development of real-time, non-destructive, and high-throughput phenotypic monitoring technology for estimating chlorophyll in rice panicles is of great significance for accurately determining the rice harvest period and synergistically increasing the annual yield of rice and wheat.

[0004] At present, there are some automatic rice ear photography devices on the market, but they can only simply take pictures of the appearance of the rice ears and cannot predict chlorophyll. Summary of the invention

[0005] In view of the above problems, the object of the present invention is to provide a method and system for estimating chlorophyll in rice ears, which can acquire rice ear images losslessly in a darkroom environment and estimate the chlorophyll content in the rice ears.

[0006] To achieve the above object, the present invention adopts the following technical solutions: In a first aspect, the present invention provides a method for estimating chlorophyll in rice panicles, comprising the following steps: Capture images of rice ear samples; After preprocessing the collected image information of the rice ear samples, the chlorophyll of the rice ear samples is predicted by inputting into a pre-established chlorophyll regression analysis model, and the maturity of the rice ear samples is estimated based on the chlorophyll prediction results.

[0007] Further, after preprocessing the collected rice ear sample image information, a pre-established chlorophyll regression analysis model is input to predict the chlorophyll of the rice ear sample, and the maturity of the rice ear sample is estimated based on the chlorophyll prediction result, including: Preprocessing the collected original images of rice ear samples; The pre-processed rice ear sample image is input into the pre-established and trained chlorophyll regression analysis model to predict the chlorophyll content of the rice ear sample; Based on the chlorophyll prediction results of rice ear samples, the maturity of rice ear samples is estimated.

[0008] Furthermore, the pre-processed rice ear sample image is input into a pre-established and trained chlorophyll regression analysis model to predict the chlorophyll of the rice ear sample, including: Collect multiple images of rice ears at different growth stages; The acquired rice canopy image data was calibrated and cropped using Python software to obtain cropped images of a preset size; Through field destructive sampling, the real chlorophyll data of rice ears that meet the preset requirements in each plot of the field are obtained; A data set was constructed based on the real chlorophyll data of rice ears, and the established chlorophyll regression analysis model was trained and tested; The collected rice ear sample images are input into the trained chlorophyll regression analysis model to predict the chlorophyll content of the rice ear samples.

[0009] Furthermore, the collecting of multiple rice ear images at different growth stages includes: setting the shooting angle of the multi-angle RGB camera to two shooting angles of 0° and 45°, and acquiring rice canopy image data in a fixed time period on every sunny day from the beginning of heading and flowering stage to the end of filling.

[0010] Further, the chlorophyll regression analysis model includes: an encoder part and a decoder part; The encoder part includes four stages for extracting multi-scale features of rice ears; The decoder part also includes four stages for gradually restoring the multi-scale features to the original image size.

[0011] Further, the encoder part includes a sequence conversion module, a plurality of Transformer modules and a first feature fusion module; The sequence conversion module is used to convert the input rice ear image into a Token sequence; Each Transformer module is used to extract features layer by layer, and each module includes a multi-head self-attention layer and a feedforward neural network; The first feature fusion module is used to fuse the multi-scale features output by each Transformer module.

[0012] Furthermore, the decoder part includes an upsampling module, a second feature fusion module and a linear projection module, wherein the upsampling module is used to upsample the multi-scale features output by the encoder part to gradually restore the spatial resolution of the rice ear image; the second feature fusion module is used to fuse features of different scales; and the linear projection module is used to map the fused features to the pixel level to obtain the classification result.

[0013] In a second aspect, the present invention provides a rice ear chlorophyll estimation system, comprising: A data acquisition device, used for acquiring images of rice ear samples; The host computer is used to pre-process the collected rice ear sample image information, input the pre-established chlorophyll regression analysis model to predict the chlorophyll of the rice ear sample, and estimate the maturity of the rice ear sample based on the chlorophyll prediction result.

[0014] In a third aspect, the present invention provides a computer-readable storage medium storing one or more programs, wherein the one or more programs include instructions, which, when executed by a computing device, cause the computing device to perform any method.

[0015] In a fourth aspect, the present invention provides a computing device, comprising: one or more processors and a memory, wherein the memory stores one or more programs and is configured to be executed by the one or more processors, and the one or more programs include instructions for executing any method.

[0016] The present invention adopts the above technical solution, which has the following advantages: 1. This paper studies a variety of deep learning model architectures and finally chooses the ViT (Vision Transformer) model as the RGB chlorophyll estimation model. The ViT model adopts an end-to-end idea to map the input data directly to the output, thus avoiding complex and subtle manual feature engineering or multi-stage processing.

[0017] 2. The present invention constructs a rice ear chlorophyll regression analysis model based on RGB images, realizing real-time and non-destructive monitoring of rice ear chlorophyll.

[0018] 3. The present invention utilizes a driving transmission device to transmit the acquired rice ear samples to a detection dark box. The shooting process of the RGB camera is completed in the detection dark box, and it has its own light source system. The monitoring is not affected by external light sources, ensuring the consistency of the monitoring environment and high monitoring stability.

[0019] Therefore, the present invention can be widely applied to the technical field of crop phenotype monitoring. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Various other advantages and benefits will become apparent to those of ordinary skill in the art by reading the detailed description of the preferred embodiments below. The accompanying drawings are only for the purpose of illustrating the preferred embodiments and are not to be considered as limiting the present invention. Throughout the accompanying drawings, the same reference numerals are used to represent the same components. In the accompanying drawings: Figure 1 is a flow chart of a method for estimating chlorophyll in rice ears provided by an embodiment of the present invention; Figure 2 This is a VIT network structure provided by an embodiment of the present invention; Figure 3 is the preliminary model result R provided by the implementation case of the present invention 2 . DETAILED DESCRIPTION

[0021] In order to make the purpose, technical solution and advantages of the embodiment of the present invention clearer, the technical solution of the embodiment of the present invention will be clearly and completely described below in conjunction with the drawings of the embodiment of the present invention. Obviously, the described embodiment is a part of the embodiment of the present invention, not all of the embodiments. Based on the described embodiment of the present invention, all other embodiments obtained by ordinary technicians in this field belong to the scope of protection of the present invention.

[0022] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit exemplary embodiments according to the present invention. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprising" and / or "including" are used in this specification, it indicates the presence of features, steps, operations, devices, components and / or combinations thereof.

[0023] In recent years, with the rapid development of technologies such as computer vision and deep learning, image segmentation based on deep learning has gradually been applied to various fields. Among them, image semantic segmentation is an image segmentation method that can classify each pixel in an image into different semantic categories. For example, in the scenario of autonomous driving, the car captures the surrounding scene in real time through the camera, uses image semantic segmentation technology to determine the predicted category of each pixel, avoids other cars and pedestrians around, or identifies lane lines to determine the driving direction; in the medical scenario, image semantic segmentation of different organs of the human body can assist doctors to better judge possible lesions in medical images. However, in the direction of rice, there are not many examples of using deep learning-based regression analysis for processing. If applied to rice, the input data is directly mapped to the output, thus avoiding complex and subtle manual feature engineering or multi-stage processing. At the same time, compared with traditional image segmentation methods, it can be more lossless and efficient, more automated and efficient.

[0024] Therefore, in some embodiments of the present invention, a method for estimating chlorophyll of rice ears is provided, the method comprising the following steps: collecting images of rice ears; after preprocessing the collected image information of rice ears, inputting a pre-established chlorophyll regression analysis model to predict the rice ears, and estimating the maturity of the rice ears based on the model. The present invention can accurately and quickly estimate the chlorophyll of rice ears by analyzing the image data of rice ears, has important practical application value, can reduce destructive detection and improve grain yield and quality.

[0025] Correspondingly, in other embodiments of the present invention, a rice ear chlorophyll estimation system, device and medium are provided.

[0026] Example 1 like Figure 1 As shown, this embodiment provides a method for estimating chlorophyll in rice ears, which comprises the following steps: 1) Collect images of rice ear samples; 2) After preprocessing the collected image information of rice ear samples, the chlorophyll of the rice ear samples is predicted by inputting the pre-established chlorophyll regression analysis model, and the maturity of the rice ear samples is estimated based on the chlorophyll prediction results.

[0027] Preferably, the above step 2) includes the following steps: 2.1) Preprocessing the collected original images of rice ear samples; 2.2) Input the pre-processed rice ear sample image into the pre-established and trained chlorophyll regression analysis model to predict the chlorophyll content of the rice ear sample; 2.3) Based on the chlorophyll prediction results of rice ear samples, the maturity of rice ear samples is estimated.

[0028] Preferably, in the above step 2.1), the collected rice ear sample image is preprocessed in order to improve the accuracy and effect of image segmentation. In this embodiment, the preprocessing mainly includes color correction and image cropping, specifically: Color correction: Based on the obtained rice ear sample image, the color of the rice ear sample image is accurately restored using the Colorchecker24 colorimetric card and the matching Colorchecker Camera Calibration software; Image cropping: Since the resolution of the original image is too large (4800×3200), in order to reduce the time consumed in the image annotation and model training process, this embodiment crops the color-corrected image into multiple cropped images with a resolution of 2016*2016.

[0029] Preferably, the above step 2.2) includes the following steps: 2.2.1) Collect multiple images of rice ears at different growth stages.

[0030] In this embodiment, when obtaining rice canopy image data, a handheld phenotyping system PhenoARM monitored by a multi-angle RGB camera is used to obtain field images. The specific shooting conditions are: the shooting angle of the multi-angle RGB camera is set to two shooting angles of 0° and 45°, and the rice canopy image data is obtained in a fixed time period on each sunny day from the heading and flowering stage to the end of the grain filling stage (i.e., the heading and flowering stage, the grain filling stage, and the maturity stage).

[0031] 2.2.2) Use Python software to calibrate and crop the acquired rice canopy image data to obtain a cropped image of a preset size.

[0032] In order to reduce the time spent in the image annotation and model training process, in this embodiment, the resolution of each rice canopy image is cropped into three cropped images of 1024*1024 size.

[0033] 2.2.3) Through field destructive sampling, the true chlorophyll values ​​of rice ears that meet the preset requirements in each plot in the field are obtained as true labels for labeling.

[0034] In this embodiment, first, a rice ear is taken from the rice ears of the varieties with good growth in each plot in the field through field destructive sampling. Then, the rice ear is threshed, shelled and treated with corresponding chemical reagents to obtain the corresponding chlorophyll true value. The method for obtaining the true value of chlorophyll in rice ears is a well-known technology for those skilled in the art, and the present invention will not elaborate on it.

[0035] 2.2.4) Construct a data set based on the real chlorophyll data of rice ears, and train and test the established chlorophyll regression analysis model.

[0036] In this embodiment, when the data set is used to train and test the chlorophyll regression analysis model, the data set needs to be randomly divided into a training set, a validation set, and a test set. The division ratio can be 7:2:1, and the present invention does not impose any limitation on this.

[0037] 2.2.5) Input the collected rice ear sample images into the trained chlorophyll regression analysis model to predict the chlorophyll content of the rice ear samples.

[0038] Furthermore, in the above step 2.2.4), in this embodiment, the chlorophyll regression analysis model adopts the VIT model. The VIT model has the following advantages: higher deployment efficiency, no need to cooperate with experts in the same field for feature engineering, and ready to use. Directly inputting the original data or only performing minimal processing on the original data may capture features that are ignored by traditional methods and perform better. Reduce the processing flow, thereby reducing error accumulation and improving accuracy. It has strong adaptability and does not require special design under different conditions (such as lighting, lens), and can automatically adapt to new data patterns during training. This series of advantages enables the ViT model to complete the task of estimating rice ear chlorophyll from rice photos.

[0039] like Figure 2 As shown in the figure, the chlorophyll regression analysis model includes: an encoder part and a decoder part. The encoder part contains four stages, each of which contains several Transformer modules to extract the multi-scale features of rice ears. The decoder part also contains four stages, and the multi-scale features are gradually restored to the original image size through the feature fusion module and the upsampling module.

[0040] Specifically, the encoder part includes a sequence conversion module, several Transformer modules and a first feature fusion module. The sequence conversion module is used to convert the input rice ear image into a token sequence; each Transformer module is used to extract features layer by layer, and each module contains a multi-head self-attention layer and a feedforward neural network; the first feature fusion module is used to fuse the multi-scale features output by each Transformer module.

[0041] The decoder part includes an upsampling module, a second feature fusion module and a linear projection module. The upsampling module is used to upsample the multi-scale features output by the encoder part and gradually restore the spatial resolution of the rice ear image; the second feature fusion module is used to fuse features of different scales; and the linear projection module is used to map the fused features to the pixel level to obtain the classification results.

[0042] The rice ear chlorophyll estimation method provided by the present invention can accurately determine the optimal harvest time and can also be used to monitor the growth status of rice ears in real time. According to the prediction of the chlorophyll estimation model, the harvesting strategy can be adjusted in time, and farmers and researchers can manage the growth and harvest of rice more accurately and efficiently.

[0043] Example 2 This embodiment takes the processing of indoor rice ear images as an example to further introduce the method of the present invention.

[0044] 1) Collect images of rice ear samples.

[0045] Rice ear images were collected, and a total of 525 indoor rice ear images were obtained, including multiple different varieties and different growth periods.

[0046] 2) After preprocessing the collected rice ear sample image information, the pre-established chlorophyll regression analysis model is input to segment the rice ear samples, and the phenotypic parameters of the rice ear samples are extracted according to the segmentation results.

[0047] The image was preprocessed, and then the Colorchecker24 colorimetric card and ColorChecker CameraCalibration software were used to accurately calibrate the image color to obtain the corrected image; since the weight of the background pixels in the original image was too large, in order to avoid the model segmenting the rice ear pixels as the background during the segmentation process, the original image was cropped and the image size after cropping was 1024×1024.

[0048] Constructing the dataset: The real chlorophyll values ​​were used as the real labels of the images in the training set and test set. The dataset obtained more than 1,200 image-to-real image pairings. The dataset was randomly divided into training set, validation set, and test set in a ratio of approximately 7:2:1.

[0049] Regression analysis model training: ① Training environment This embodiment is based on the Ubuntu 20.04.5 LTS system, the GPU is NVIDIA Quadro RTX8000, the CPU is Intel(R) Xeon(R) Silver 4215 CPU @ 2.50GHz × 32, the image processing language is Python3.7, tensorflow-gpu1.14.0.

[0050] ② Training parameter setting In this embodiment, the training samples from the training set are input into the network by batch training. The batch size of the model training set and the validation set is set to 4, that is, each batch includes 4 samples, and 151 batches are required to traverse the data set in the field environment, and 102 batches are required to traverse the data set in the indoor environment. The number of training network iterations is set to 100 rounds, the initial learning rate of the model is set to 0.001, and the others are set to default parameters. And the validation loss is monitored by setting the "early_stopping" module, that is, when the loss of the validation set decreases less than the learning rate of 0.001, the training stops, in order to prevent the model from overfitting.

[0051] Model evaluation: In order to verify whether the image segmented by the model can be used for subsequent ear feature extraction, it is necessary to accurately measure the segmentation performance of the model. The model determination coefficient R 2 Evaluate the model's predictive performance.

[0052] ①Model determination coefficient R 2 Accuracy refers to the proportion of correctly predicted samples to all samples, and the calculation formula is as follows:

[0053] in, is the target variable value predicted using the regression model; is the true value of the sample target variable.

[0054] The results of the target variable value predicted by the regression model and the true value of the sample target variable are as follows Figure 3 shown.

[0055] 3) The chlorophyll of the rice ear samples is estimated based on the phenotypic parameters of the rice ear samples, and the maturity of the rice ear samples is estimated based on the chlorophyll, so as to achieve a more accurate and efficient prediction of the rice harvest time.

[0056] Chlorophyll model establishment and training: This embodiment is based on the Ubuntu 20.04.5 LTS system, the GPU is NVIDIA Quadro RTX8000, the CPU is Intel(R) Xeon(R) Silver 4215 CPU @ 2.50GHz × 32, the image processing language is Python3.7, tensorflow-gpu1.14.0.

[0057] The coefficient of determination R 2 The three evaluation indicators of RMSE, root mean square error (RMSE) and mean absolute error (MAE) were used to evaluate the fitting degree between the predicted value and the measured value of the rice ear chlorophyll monitoring model.

[0058] This research result suggests that the rich color and texture feature information contained in RGB images can reflect the changes in physiological indicators of crop organs, plants and populations, and can accurately reproduce the growth and development conditions of crops, showing that crop phenotyping research based on RGB images has broad application prospects.

[0059] Example 3 The above-mentioned embodiment 1 provides a rice ear chlorophyll estimation method, and correspondingly, this embodiment provides a rice ear chlorophyll estimation system. The system provided in this embodiment can implement the rice ear chlorophyll estimation method of embodiment 1, and the system can be implemented by software, hardware, or a combination of software and hardware. For example, the system may include integrated or separate functional modules or functional units to execute the corresponding steps in each method of embodiment 1. Since the system of this embodiment is basically similar to the method embodiment, the process described in this embodiment is relatively simple, and the relevant parts can refer to the partial description of embodiment 1. The embodiment of the system provided in this embodiment is only schematic.

[0060] This embodiment provides a rice ear chlorophyll estimation system, which includes: A data acquisition device, used for acquiring images of rice ear samples; The host computer is used to pre-process the collected rice ear sample image information, input the pre-established chlorophyll regression analysis model to predict the chlorophyll of the rice ear sample, and estimate the maturity of the rice ear sample based on the chlorophyll prediction result.

[0061] Preferably, the data acquisition device includes a drive transmission device and an image acquisition device. The drive transmission device is used to transmit the collected rice ear samples to the image acquisition device; the image acquisition device is used to collect images of the rice ear samples and send them to the host computer; the host computer is used to control the drive transmission device and the image acquisition device, and at the same time process the collected images to obtain the phenotypic parameters of the rice ear samples, and estimate the chlorophyll of the rice ear samples based on the phenotypic parameters of the rice ear samples.

[0062] Preferably, the driving and conveying device includes a roller conveyor and a PLC controller. The roller conveyor is arranged on a frame, and includes a conveyor belt and a stepper motor, and the stepper motor is connected to the PLC controller. The surface of the conveyor belt is provided with anti-slip material, and is used to convey the rice ear samples and ensure that the rice ear samples do not deviate during the conveying process; the PLC controller is used to control the running state of the conveyor belt through the motor according to the control signal sent by the host computer.

[0063] Preferably, the image acquisition device includes a darkroom, a calibration plate, a lighting system, an RGB camera and a displacement and photoelectric sensor; automatic doors and photoelectric sensors are respectively arranged at positions corresponding to the entrance and exit of the conveyor belt in the darkroom, and the automatic doors and photoelectric sensors are used to detect the position of the rice ear sample. When the photoelectric sensor detects that the rice ear sample has arrived at the designated position, it sends a signal to the host computer, and the host computer controls the opening and closing of the automatic door; a high-precision displacement sensor is also arranged at a preset position inside the darkroom, and when the displacement sensor detects that the rice ear sample has arrived at the designated position, it sends a signal to the host computer, and the host computer controls the RGB camera arranged at the center of the top of the darkroom to capture images of the rice ear sample; a high-light-absorbing material is also arranged on the inner wall of the darkroom to form a stable imaging environment; the calibration plate is used to calibrate the imaging system and obtain accurate RGB values; the lighting system includes high-color rendering LED light sources evenly distributed on the top and around the darkroom to provide stable and uniform lighting.

[0064] Preferably, in this embodiment, the RGB camera uses a Sony RX0 camera, which uses a high-pixel 1-inch back-illuminated CMOS sensor and is equipped with a powerful image processor to capture high-quality rice ear images. The Sony RX0 camera has an effective pixel count of 15.3 million and a resolution of 4800x3200, which can meet the use requirements of a small darkroom space. The lens is a fixed-focal-length 24mm Zeiss Tessar T* lens, which can capture wide-angle images and has good optical performance and color reproduction capabilities.

[0065] Preferably, the system can also be remotely controlled and managed through the network, and can be automatically processed and managed according to different needs, reducing manual intervention and improving work efficiency.

[0066] Preferably, the system can also be equipped with a human-computer interaction interface in the host computer, which can intuitively display the image of the rice ear sample and the extracted phenotypic information, making it convenient for users to monitor and manage.

[0067] When the system is in use, the process is: ① Set the speed and start and stop time of the conveyor belt (combined with the photo taking time setting) through the controller, and arrange the rice ear samples with the roots cut off and place them on the conveyor belt; when collecting rice ear samples, rice samples can be taken from a row of the field to be tested, and the rice samples can be cut to retain only the rice ear part; ② Start the motor, and the conveyor belt rotates to transport the rice ear samples to the inside of the detection dark box; ③ When the rice ear sample enters the detection dark box (the color card is placed in the upper left corner of the dark box for taking photos), the third sensor starts working. If it detects that the rice ear sample reaches the bottom of the RGB camera, it sends a signal to the host computer; ④ The host computer controls the RGB camera, which takes pictures of the rice ear samples and transmits the collected images to the host computer, which numbers and stores the rice ears of the rice ear samples; Repeat the above steps ① to ④ until the data collection of all rice ear samples is completed; The above are the detailed steps for implementing the device, which can be adjusted according to specific circumstances. In addition, attention should be paid to safety issues during implementation to ensure that the device can operate stably.

[0068] Example 4 This embodiment provides a processing device corresponding to the rice ear chlorophyll estimation method provided in this embodiment 1. The processing device can be a processing device for a client, such as a mobile phone, a laptop computer, a tablet computer, a desktop computer, etc., to execute the method of embodiment 1.

[0069] The processing device includes a processor, a memory, a communication interface and a bus, and the processor, the memory and the communication interface are connected through the bus to complete mutual communication. The memory stores a computer program that can be run on the processor, and the processor executes the rice ear chlorophyll estimation method provided in this embodiment 1 when running the computer program.

[0070] Preferably, the memory may be a high-speed random access memory (RAM), and may also include a non-volatile memory, such as at least one disk memory.

[0071] Preferably, the processor may be a central processing unit (CPU), a digital signal processor (DSP) or other general-purpose processors of various types, which are not limited here.

[0072] Example 5 The rice ear chlorophyll estimation method of this embodiment 1 can be specifically implemented as a computer program product, which may include a computer-readable storage medium carrying computer-readable program instructions for executing the rice ear chlorophyll estimation method based on deep learning described in this embodiment 1.

[0073] Computer readable storage media can be tangible devices that hold and store instructions used by instruction execution devices. Computer readable storage media can be, for example, but not limited to, electronic storage devices, magnetic storage devices, optical storage devices, electromagnetic storage devices, semiconductor storage devices, or any combination thereof.

[0074] It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0075] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0076] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 One or more processes and / or boxes Figure 1 A function specified in one or more boxes.

[0077] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0078] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A method for estimating chlorophyll in rice ears, characterized in that: The following steps are involved: Capture images of rice ear samples; After preprocessing the collected rice ear sample image information, the pre-established chlorophyll regression analysis model is input to predict the chlorophyll of the rice ear sample, and the maturity of the rice ear sample is estimated based on the chlorophyll prediction result; After preprocessing the collected rice ear sample image information, the pre-established chlorophyll regression analysis model is input to predict the chlorophyll of the rice ear sample, and the maturity of the rice ear sample is estimated based on the chlorophyll prediction result, including: Preprocessing the collected original images of rice ear samples; The pre-processed rice ear sample image is input into the pre-established and trained chlorophyll regression analysis model to predict the chlorophyll of the rice ear sample; The pre-processed rice ear sample image is input into a pre-established and trained chlorophyll regression analysis model to predict the chlorophyll of the rice ear sample, including: Based on the chlorophyll prediction results of rice ear samples, the maturity of rice ear samples is estimated; Collect multiple images of rice ears at different growth stages; The acquired rice canopy image data was calibrated and cropped using Python software to obtain cropped images of a preset size; Through field destructive sampling, the real chlorophyll data of rice ears that meet the preset requirements in each plot of the field are obtained; A data set was constructed based on the real chlorophyll data of rice ears, and the established chlorophyll regression analysis model was trained and tested; Input the collected rice ear sample images into the trained chlorophyll regression analysis model to predict the chlorophyll content of the rice ear samples; The collecting of multiple rice ear images at different growth stages includes: setting the shooting angle of the multi-angle RGB camera to two shooting angles of 0° and 45°, and acquiring rice canopy image data in a fixed time period on each sunny day from the beginning of heading and flowering to the end of grain filling; The chlorophyll regression analysis model adopts the ViT model, which includes an encoder part and a decoder part; The encoder part includes four stages for extracting multi-scale features of rice ears; The decoder part also includes four stages, which are used to gradually restore the multi-scale features to the original image size to obtain the classification results.

2. A rice ear chlorophyll estimation method as claimed in claim 1, characterized in that: The encoder part includes a sequence conversion module, a plurality of Transformer modules and a first feature fusion module; The sequence conversion module is used to convert the input rice ear image into a Token sequence; Each Transformer module is used to extract features layer by layer, and each module includes a multi-head self-attention layer and a feedforward neural network; The first feature fusion module is used to fuse the multi-scale features output by each Transformer module.

3. A rice ear chlorophyll estimation method as claimed in claim 1, characterized in that: The decoder part includes an upsampling module, a second feature fusion module and a linear projection module, wherein the upsampling module is used to upsample the multi-scale features output by the encoder part to gradually restore the spatial resolution of the rice ear image; the second feature fusion module is used to fuse features of different scales; and the linear projection module is used to map the fused features to the pixel level to obtain the classification result.

4. A rice ear chlorophyll estimation system, characterized in that: include: A data acquisition device, used for acquiring images of rice ear samples; The host computer is used to pre-process the collected rice ear sample image information, input the pre-established chlorophyll regression analysis model to predict the chlorophyll of the rice ear sample, and estimate the maturity of the rice ear sample based on the chlorophyll prediction result; After preprocessing the collected rice ear sample image information, the pre-established chlorophyll regression analysis model is input to predict the chlorophyll of the rice ear sample, and the maturity of the rice ear sample is estimated based on the chlorophyll prediction result, including: Preprocessing the collected original images of rice ear samples; The pre-processed rice ear sample image is input into the pre-established and trained chlorophyll regression analysis model to predict the chlorophyll of the rice ear sample; The pre-processed rice ear sample image is input into a pre-established and trained chlorophyll regression analysis model to predict the chlorophyll of the rice ear sample, including: Based on the chlorophyll prediction results of rice ear samples, the maturity of rice ear samples is estimated; Collect multiple images of rice ears at different growth stages; The acquired rice canopy image data was calibrated and cropped using Python software to obtain cropped images of a preset size; Through field destructive sampling, the real chlorophyll data of rice ears that meet the preset requirements in each plot of the field are obtained; A data set was constructed based on the real chlorophyll data of rice ears, and the established chlorophyll regression analysis model was trained and tested; Input the collected rice ear sample images into the trained chlorophyll regression analysis model to predict the chlorophyll content of the rice ear samples; The collecting of multiple rice ear images at different growth stages includes: setting the shooting angle of the multi-angle RGB camera to two shooting angles of 0° and 45°, and acquiring rice canopy image data in a fixed time period on each sunny day from the beginning of heading and flowering to the end of grain filling; The chlorophyll regression analysis model adopts the ViT model, which includes an encoder part and a decoder part; The encoder part includes four stages for extracting multi-scale features of rice ears; The decoder part also includes four stages, which are used to gradually restore the multi-scale features to the original image size to obtain the classification results.

5. A computer-readable storage medium storing one or more programs, characterized in that: The one or more programs include instructions, which, when executed by a computing device, cause the computing device to perform any one of the methods of claims 1 to 3.

6. A computing device, characterized in that include: One or more processors and a memory, wherein the memory stores one or more programs and is configured to be executed by the one or more processors, wherein the one or more programs include instructions for executing any one of the methods described in claims 1 to 3.

Images