Research Of Crop Growth Status Recognition Algorithm Based On Image Technology

The observation of crop growth status is a vital part of agricultural meteorological observations,while the traditional method mainly depends on manual observation.The observation method is complex,subjectively influenced and inefficient.With the mature application of sensor detection technology and remote network transmission technology,crop observations gradually shift from manual observation to automated observation.Moreover,the automatic observation of crops such as cotton,corn,rice,and wheat is exactly necessary in some major project.In this paper,automatic observation of corn growth status in different fields in the field can be realized through the establishment of an automatic observation station for crop meteorology under the above background.1.Proposing a crop segmentation algorithm based on multi-threshold parameters.Through analyzing the lighting characteristics of real crop images,a multi-threshold parameter green vegetation segmentation method was proposed.This method can reduce the impact of dramatic changes in light intensity on the segmentation results under conditions of large dynamic range of light intensity.Comparing with the segmentation results of EXG,CIVE and AP-HI,it could be concluded that this method is more ideal for segmentation results under specific circumstances.On this basis,the automatic recognition of daily corn plant coverage was accomplished by setting the daytime image coverage change rate δ and the daily coverage change rate ρ.2.In this paper,the laser range finder is innovatively applied to the field of crop canopy height measurement.It has eliminated many defects such as the high cost of crop canopy height measurement based on binocular vision and mechanized operations of monocular vision measurement methods.In this paper,the laser distance measuring device is installed in a three-dimensional head,and the dynamic scanning of the observation area is achieved by controlling the rotation of the three-dimensional head to obtain the distance set of the emission point and the measured point.Furthermore,automatic measurement of crop canopy height comes true through mathematical modeling and error analysis.The application results show that the device can automatically measure the height of the canopy and then obtain the exact height of the plant height.The measurement error is ±3cm.3.According to the seedling emergence characteristics of corn,establish a geometric model of corn seedlings.Using the proportional relationship of the model,extract the typical target area in the segmentation result,calculate the spatial distribution of the target to obtain image emergence information,combine daily image data,and calculate daily emergence probability to complete automatic identification of corn seedlings.The identification of the seven-leaf stage uses a comprehensive judgment of the rate of change of the individual seedlings’ pixels and the canopy height of the plant,which reduces the error caused by relying only on pixel judgments.4.Using machine learning method to realize automatic identification of corp.According to the different types of corn in different locations to extract the different characteristics of the typical target characteristics,the establishment of a relatively complete characteristics of the characteristics of the library learning,improve the tassel classification model.By comparison with the signature database,automatic identification of the tasseling period of the corn is achieved.