| Masson pine is distributed in southern China and is a pioneer tree species for barren mountain afforestation.The quality of masson pine seedlings is related to the survival rate of afforestation.Generally,the indicators for evaluating the quality of seedlings include morphological indicators,physiological indicators and dynamic indicators.The traditional method of measuring the above indicators is cumbersome,time-consuming,laborious,lacking in accuracy,and will damage the seedlings.Therefore,the research on efficient,accurate and nondestructive seedling index measurement methods is particularly important for the rapid evaluation of seedling quality and selection of high-quality seedlings.This article aims to carry out research on masson pine vitality detection technology through multi-information fusion to improve the accuracy of monitoring.Using multi-scale shortcut neural network,laser speckle technology and YOLO network model,the physiological indicators(water content and nitrogen content),root growth rate indicators,and morphological indicators(crown,trunk and roots)of Masson pine seedlings were quickly analyzed.Accurate measurement.The main contents are as follows:(1)Using a multi-scale shortcut convolutional neural network to quickly,accurately and non-destructively detect the leaf moisture and nitrogen content of Masson pine seedlings.Constructed a multi-scale shortcut convolutional neural network,using near-infrared hyperspectral data to predict the water content and nitrogen content of Masson pine seedling leaves.By designing a reasonable shortcut structure,the information loss caused by multi-layer transmission during forward propagation is reduced.At the same time,the loss caused by error multi-layer transmission during back propagation is reduced,and the gradient disappearance during training is avoided,so as to obtain accuracy.Measurement accuracy.The experimental results show that the performance of the proposed model in predicting the leaf water content and nitrogen content of Masson pine seedlings is the best compared with other commonly used models.In the test of masson pine seedling leaf water content,the measured value on the test set The correlation coefficient with the true value is as high as 0.977,and the measurement root mean square error is 0.242;in the masson pine seedling leaf nitrogen content test,the correlation coefficient between the measured value and the true value on the test set is 0.906,and the measurement root mean square error is 0.061.(2)Using laser speckle method to measure the tiny morphological changes of the roots of Masson pine.Using statistical interferometry to derive the probability density function of the speckle phase,and measuring the displacement of the roots of the seedlings in a short period of time through the phase of the object,reducing the time to measure the growth potential of the seedlings’ roots from the traditional methods of several months to less than a few seconds.,Which greatly improves the measurement efficiency.In order to verify the validity of the measurement method,a water stress experiment on seedlings was designed to obtain different changes in the root morphology of seedlings when water is sufficient and deficient.In the experiment,water stress was formed by reducing the water content of some masson pine to affect the vitality of masson pine,and masson pine was divided into two groups according to whether it was subjected to water stress.By measuring the root elongation of each seedling,it is found that the root elongation can reflect the vitality of the seedlings and is more efficient than traditional methods.(3)Using the YOLO network model to measure the morphological indicators of Masson pine seedlings.Acquire images through Real Sense depth camera,make the image data into a data set and use YOLO for feature extraction.After verifying the validity of the model,the accurate extraction of masson pine seedling morphological indicators,including the crown,trunk and tree,is realized from the sample image.The height of the root.In the experiment,the morphological parameters extracted by the model are compared with the data measured manually using a ruler,and the error is within 5%.Therefore,the measurement result of the network model is accurate and reliable,which can replace manual measurement to a certain extent.The experimental system studied in this paper is stable,accurate and reliable in measuring the physiological indexes,morphological indexes and root elongation of Masson pine seedlings,which can provide accurate data basis for the cultivation and management of Masson pine seedlings. |
