Research On Insect Monitoring System Based On Android

Forestry is the basic industry for ecological civilization.There are so many problems about forestry in China,such as backward production equipment and weak disaster resistance etc.In recent years the biggest problem for Chinese forestry is insect pest,traditional insect pest monitoring is based on manual counting.This traditional method is inefficiency and cost a large number of manpower and material resources at the same time.So people need a modern insect pest monitoring technology.With the rapid development of image processing,image processing has become the most effective way for insect counting.Most of insect pest monitoring systems based on image processing are used to install industrial camera on trap lamp to collect and transmit images to computer.With the rapid development of Android smartphones,in this paper,the image processing algorithm has been transplanted to smart phone and the whole insect pest monitoring system has been completed on Android platform.The main research contents and results are as follows:(1)System pays attention to the problem of unbalanced illumination in the field and takes Multi-scale Retinex as the solution.Aiming at the problem of deformation image caused by camera angle,perspective transformation has been used to complete image correction,which eliminates the error caused by camera angle and improves the accuracy of insect counting.(2)An improved L0 norm algorithm based on four direction gradient operator was presented,aiming to the problem about insect extraction caused by background texture and rough surface,this algorithm eliminate the background texture effectively while more details of insect has been remained.(3)Developed an insect pest monitoring application based on Android platform,which transplanted image preprocessing algorithm and insect counting algorithm into mobile phone to complete insect counting.Image processing unit use C/C++and Java to compile,which is more effective.The system has integrated map localization and communication to server and aliyun OSS.In order to verify the accuracy of insect counting,using this system to monitor Sanqingyuan garden in Haidian District of Beijing.According to experimental results,the accuracy of insect counting is about 88.75%,and the average time consumption is 6.487 seconds,which meets the requirements of counting accuracy and processing speed in practical work.System has lower hardware cost and easy manipulation.It is suitable for popularization among junior forestry workers.