The Research On The Embedded System Of Image Acquisition And Diagnosis For Rice Disease And Pest

At present,China's forecasting work of rice disease and pest is backward and mainly rely on experience of grass-roots forecasting staff.There are problems such as low efficiency,subjectivity and non-real-time,which could not meet the current demand of high incidence of rice disease and pest.In order to realize the intelligent forecast of rice disease and pest,some scholars have begun to study the application of image-based intelligent identification and diagnosis technology of rice disease and pest in disease and pest forecasting,the current agricultural image acquisition tools are not convenient and immovable,or difficult to collect the specified disease and pest areas which is hard to reach by hands and sight.In this paper,we use wireless network,embedded and image processing technology to design and establish the embedded system of image acquisition and diagnosis for rice agricultural disease and pest.The main contents and results of this paper are included:(1)Design and establish an embedded rice disease and pest image acquisition equipment which includes an embedded wireless camera,a hand-held pole and a handset console.The embedded wireless camera module: Using V4L2 programming interface to capture the original YUV image and to generate a single image file through JPEG encoding;The original image data is encoded by H.264 and the video stream is transmitted by RTP/RTSP streaming protocol; Using Socket programming to establish TCP/UDP server and to achieve double-ended control communication through custom protocol interface.Mobile phone control module: Building RTP/RTSP streaming media client with VLC open source library to realize RTSP session with embedded camera and decoding and playback of RTP data;Establishing TCP/UDP client with Socket programming as the server to achieve the sending of control command and the receiving of return command according to the communication interface;Selecting the image asynchronously strategy to realize a preview of the captured image and integrating upload and local image processing functions.(2)Study and realize the diagnose algorithm of rice sheath blight that includes image pre-processing,background segmentation,feature extraction,recognition and diagnosis: Pre-processing the image of sheath blight disease and combining the super-color space and the significant segmentation to separate the lesion area; The color,shape and texture features of each spot were extracted;Using SVM classifier to identify sheath blight disease spot and giving the damage level according to the area proportion of sheath blight disease spot.In this paper,the embedded system of image acquisition and diagnosis for rice agricultural disease and pest can be used to collect the image of rice disease and pest conveniently and realize the intelligent diagnosis of rice sheath blight.The stretchable pole could send the camera to the specified agricultural disease and pest areas which is hard to reach by hands and sight.The mobile phone could preview the scene continuously from the front-end camera and control the camera.Mobile phone control terminal can choose an original image or pre-processed feature data to upload to the image processing server for sheath blight disease identification and diagnosis.The system has good application prospects and can be widely used in rice field survey and forecasting of disease and pest.