Developing real-time detecting system for apple color and size based on ARM

Abstract. Size and color are two important indexes to measure the external quality of apples. The current apple size and color detection methods are mostly designed on PC platform, which are difficult to be popularized to smaller companies and growers for the high expense. This research designed and implemented a novel detection system based on embedded platform that covers full function of apple size and color detection but with much lower cost. The hardware of system was mainly composed of ARM11 processor, CMOS camera and stepping motor. The real-time image of apple was acquired by CMOS camera and diameter value was extracted from image as index of size after image preprocessing. Considering that HSI color model is more close to human visual perception than RGB color model, the RGB image acquired by the CMOS camera was transformed to HSI color model. The H component was calculated to determine the percents of red component, which was used for color grading of apple. Preliminary experiments show that the average detection time is 300ms, color detection accuracy is 90% and size (60×60 pixel) detection accuracy is 93%, respectively. Compared with the traditional apple detection system, this system has the advantages of higher detection speed, lower cost, and smaller size. It will have a good prospect of application in apple quality detecting areas.