A Memory Efficient Algorithm for Real Time Object Counting
S. Subramanian, L. V. Bhadrinarayana · 2009
Object Counting is a challenging problem with different solutions based on the available computing power and the nature of data to be processed. Memory efficiency, simplicity and speed are very important requisites for algorithms to be used in modern day systems which include distributed systems and wireless sensor networks where it is extremely advantageous to do basic preprocessing of data from the sensors in the nodes itself. Reduced computing power available in the nodes poses a challenge and this can be overcome by the use of algorithms with simple steps. We present an algorithm to monitor continuity, count objects and measure parameters such as area by isolating patterns and objects from data based on simple computations. Memory efficiency, speed and flexibility of the proposed algorithm have been discussed. It has many broad applications that include counting objects on a conveyer, monitoring people in a traffic signal, pattern isolation from multicolored images and monitoring continuity in real time image data from sensors. We describe the experimental setup used to implement this algorithm.