In the paper a novel harmony search (HS) algorithm based on opposition and differential evolution (ODHS) algorithm is proposed in order to solve high dimensional optimization problems. It provides a new architecture of hybrid algorithms, which organically merges the differential evolution (DE) into HS algorithm and the ODHS algorithm initializes the HM (harmony memory) using opposition based learning and uses “opposites” selection replacing random selection. During the course of evolvement, harmony search and differential evolution is alternately used to improve the search performance, which makes the ODHS algorithm have more powerful exploitation capabilities. Simulation and comparisons based on four benchmark functions demonstrate the effectiveness, efficiency and robustness of the proposed ODHS.