Abstract:

Currently,most of the rice disease image recognition systems rely on digital cameras and computers,which lack of portability and realtime features.To diagnosis rice diseases in the fields timely,we design a rice disease image recognition system based on Android mobile phones.We analyze the color,shape,and texture features of the following four types of diseases:rice blast, flax spot disease,dry pointed nematode,and white leaves dry disease.We use multiple methods such as image pretreatment,image segmentation,feature extraction and image recognition.Experimental results show that the accuracy rate is 93.7%,the detection rate is 96.22%,the false detection rate is 6.22%,the false alarm rate is 1.56%,and the average diagnosis time is 2.802 s.The proposed system can effectively capture and diagnose the types of rice disease rapidly and accurately,and it can also provide some disease control measures.

Key words: Android mobile phone;rice disease;diagnosis;image recognition