The goal of this study was to develop a method based on Android OS mobile phone by using image processing to rapidly and non-destructively determine leaf area (LA). Software using Java language was used to estimate LA. The steps for processing each image were as follows: image collection, image segmentation, image binarazition, filtering and LA calculation. The proposed method was carried out on triangle of 100mm2, square of 100mm2, circle of 314.15mm2 and five-pointed star of 112.26mm2 in eight distance ranges by AutoCAD precision drawings. The test result showed that the measured error was in the range of -0.62% to 0.79%. The proposed method was applied to measure different leaves representing varied shapes and sizes, such as tomato, eggplant and maple representing. The measured error was within ±1% compared with type LI—3100 leaf measuring device when the distance between mobile and leaf was 300～600mm. The picture sizes were 2560 pixels×1920 pixels and the measuring accuracy was 0.001cm2 in this work, which showed the proposed method could measure leaf areas accurately.