One exciting prospect of modern construction projects is the potential for multimedia techniques, such as real-time video, to significantly affect the way of project delivery. Among all the techniques in real-time video processing, template-matching plays the most essential role because of its high computational complexity and its ability to deal with considerable redundancy.
However, commonly used template-matching techniques in spatial domain cannot meet all of the requirements for all construction applications. Some methods have heavy computational burden, others suffer from inadequate accuracy. Therefore, an adjustable template-matching capable of meeting all requirements is an exciting prospect.
The proposed template-matching algorithm utilizes special relation between associated Hadamard determinants. Results indicate the proposed algorithm outperforms many popular algorithms without increasing computational complexity level. Moreover, the algorithm is capable of adjusting three parameters accordingly to meet different construction-related applications.
Source: University of Maryland
Author: Yang, Xue