| Abstract: | Abstract: In order to achieve fast real-time identification and diagnosis of field pests, a remote automatic pest identification system was designed in this paper. This system is composed of remote classification platform (ROCP) including personal computer, CMOS camera and 3G wireless communication module and a host control platform (HCP). The ROCP sends the image data, which is encoded using JPEG 2000, to the HCP through the 3G network. The image transmission and communication are accomplished using 3G technology. The system transmits the data via a commercial base station. The system can work properly based on the effective coverage of base stations, no matter the distance from the ROCP to the HCP. The image data was decoded firstly, then the pest was segmented from background, and the morphology features and color features were extracted at last for classification. Sixteen morphology features consisted of perimeter, area, eccentricity and seven Hu invariant moments etc. Nine color features were described by color moments. The support vector machine classifier was used at last for identification. Six species of common field pests including Cnaphalocrocis medinalis Guenee, Chilo suppressalis, Sesamia inferens, Naranga aenesc, Anomala corpulenta Motschulsky, Prodenia litura were tested in the system and the average accuracy is 87.4%. Considering the different pests' pose and different field lighting conditions, the result is satisfactory. The study of the automatic pest identification system which combined with machine vision, image processing, pattern recognition technology and 3G wireless communication technology, was not reported in China. The designed system can automatically identify the field pests and can provide timely and accurate information for pest prevention. The application of the designed system can reduce prevention cost and improve the control effect. The study can provide a reference for agricultural pest prevention. |
