Optimization of blood vessel detection in retina images using multithreading and native code for portable devices

Due to the importance of blood vessel detection in many medical tools and the increasing demand for portable diagnosis equipment, fast blood vessel detection algorithm in a standalone and portable device is very important. The optimization of a computationally intensive algorithm such as this on a mobile platform is challenging due to the limited resources available. In this paper, the blood vessel detection system is implemented and optimized in a portable device running Android OS on an ARM-based processor. The performance of Java programming model and native programming model are compared with respect to the execution time for blood vessel detection. The experimental results show that the blood vessel detection system has worked well in the ARM platform with the Android OS. Moreover, native programming platform is faster than Java programming with 75.91% better in terms of execution time on average. With multithreading, the performance gain in native programming is 92.77% faster than Java programming.