Research On The Embedded System Of Image Acquisition And Processing For Rice Light-trap Pests

Nowadays, black light lamps are widely used to trap insects in paddy field for monitoring rice pest population dynamics. These trapped insects are brought back the next day. The insects which don’t damage rice are removed and only rice pests are identified and counted by plant protection technicians. However, artificial identification and counting of rice pests is time-consuming and labor-exhausting,which brings about some problems, such as low identification accurate rate, low counting accuracy and non-realtime. In the paper, based on the embedded control technology, a new image acquisition and processing system for rice light-trap pests is designed to automatically achieve image acquiring and feature extraction, which are taken as the source datasets of automatic identification and counting in the future. The main contributions of the paper include:(1) An insect image acquisition module for rice light-trap pests is designed and implemented based on A80 Optimus platform. The original images are acquired by V4L2, then encoded to JPEG and finally written with GPS according to EXIF standard. This module can acquire the rice light-trap insect images with high resolution in real time.(2) An embedded image processing technology for rice light-trap insects is studied.To reduce the expense of internet access and mitigate the pressure of server, the image algorithms are transplanted into the embedded device in the work. First,Open CV is transplanted into ARM development board as an image processing library. Then, after the images of rice light-trap insect are preprocessed and marked by the connected analysis, a single insect image is obtained. Finally, 32 features of each light-trap insect such as color, morphology and textures are extracted to formulate the agricultural information unit that is sent to server for the identification and counting pests.(3) An image information transmission module for rice light-trap insects is designed and implemented. First, a C/S based structure is developed our rice light-trap system and the module uses Huawei EC1260 to access 3G wireless. Second, the formats of transfer instruction and agricultural information unit are specified by the efficient and reliable physical layer protocol which is customized by the module. Finally, the server controls access to the light-trap device using its commands.