Research On Moisture Content Detection Of Lettuce Leaves Based On Image Processing And Implementation Of Android Platform

Water content affects the physiological metabolism and growth of crop leaves directly.It is an essential index in the process of healthy growth of crops.It is very important to realize scientific,rapid and accurate non-destructive measurement of leaf water content for real-time understanding of crop moisture and guiding irrigation.In recent years,the nondestructive testing methods of crop moisture content mainly focus on the spectral technology and computer vision technology.The spectral technique adopts point sampling method,which can not represent the whole sample information well,so it will lead to strong randomness of the detection results.Computer vision technology can collect image information of each part of the target object,and can overcome the defects of sampling point information,strict sampling requirements and small sampling range in the process of spectral analysis.When the water content of crop leaves is detected by computer vision,the effect of image segmentation will be affected by lens reflection.Therefore,an image segmentation method based on PSO+Otsu(S)algorithm was proposed in this paper.In addition,with the rapid popularization and application of Android technology,more and more functions can be implemented in the Android system.Therefore,this paper put forward a method based on Android platform to detect the moisture content of lettuce leaves.The main contents and conclusions are as follows:(1)In the greenhouse,soilless cultivation techniques were used to cultivate lettuce samples.Under the condition of maintaining the same nutrient solution,four levels of water were applied to cultivate lettuce leaves with a certain moisture content gradient.These work lay a good foundation for the following research.(2)The pretreatment of lettuce leaf images was carried out,including image grayscale,denoising,segmentation and morphological processing.In order to solve the problem of uneven image brightness caused by lens reflection,a new PSO+Otsu(S)segmentation algorithm was proposed,and the algorithm was compared with the traditional Otsu algorithm.The results showed that PSO+Otsu(S)algorithm had better segmentation effect,faster algorithm implementation,and the program running time was about 1/2 of Otsu algorithm.(3)Twenty-six characteristic parameters related to the moisture content of lettuce leaves were extracted from three angles of color,shape and texture,and twenty-six characteristic parameters were screened by Pearson correlation coefficient method.The results showed that when the correlation coefficient was less than 0.01,a total of seven image characteristics related to the moisture content of lettuce leaves were screened out;when the correlation coefficient was less than 0.05,a total of twelve image characteristics related to the moisture content of lettuce leaves were screened out.(4)Using multiple linear regression(MLR),partial least squares regression(PLSR)and support vector regression(SVR)of three kinds of mathematical modeling method to establish the relation model between the moisture content of lettuce leaves and the image characteristic parameters,each model respectively on different feature modeling.The decision coefficient R~2 and the root mean square error RMSE were chosen as the model performance criteria.The results showed that the prediction ability of PLSR model based on seven characteristic parameters was the best,the test set determination coefficient R~2 was 0.902,and the root mean square error RMSE was0.302.(5)According to the demand analysis of the moisture content detection system of lettuce leaves based on Android platform,the software was divided into 5 functional modules.For each module design interface and implementation algorithm,set up three Button components,an ImageView component,a number of TextView components,and used the XML layout file to write the interface.The multiple linear regressionmodel Y=B*0.377-H*185.136+S*23.639+r*86.801+g*102.906+h*145.95+s*52.246-94.315between the characteristic parameters and water content of lettuce leaves was established,and the characteristic data was brought into the model.Finally,the water content detection function of lettuce leaves based on Android platform was realized.