Efficient implementation of morphological opening and closing by reconstruction on multi-core parallel systems

We present an efficient parallel implementation of morphological opening and closing by reconstruction operations, which have been used in the past for extracting relevant features prior to classification of remotely sensed hyperspectral images using morphological profiles. The proposed implementation has been developed and tested on various multi-core parallel platforms. These types of multi-processor systems are increasingly being used as a commodity parallel computing platform in different application domains. Our experimental results demonstrate that the proposed parallel codes fully exploit the processing power available in the considered multi-core machines.