The Research And Design Of An Embedded Object Detection System Based On Android And OpenCV

With the development of economy,people have higher requirements on material life and spiritual life.As human costs rising,consumers have higher demands on products and service quality in the fields of production,life and service.It is a matter of success and failure for enterprises,especially for manufacturing.In order to solve the above problems,more and more manufacturing enterprises want to use industrial vision system instead of traditional manual recognition to improve product quality and significantly reduce labor costs in anywhere.Nowadays,industrial vision system is expensive to purchase,maintain and operate.There is not a universal and image recognition solution for components' feature detection on the assembly lines.This paper has implemented a low-cost and scalable embedded object detection system by using Android system based on ARM Cortex-A8 platform.This system used JNI to realize the peripheral control function,supported the recognition algorithm based on the C++ inter-face of OpenCV,accomplished the independent USB Camera video capture function by V4L2 and used Android App as the user UI interface.To identify the both sides of the specified object,an ap-proach based on projection principle and frames differencing was proposed and optimized.The pro-posed algorithm is used to verify the practicability of the object detection system in industrial field.Experimental results show that the functional components of this system run well and the recognition accuracy rate has reached 99.8%,detection rate is faster than 1/s.The designed system can meet in-dustrial applications needs well.