In order to discover and employ more spectrum opportunities for the secondary users in cognitive radio networks, a multi-channel spectrum sensing period optimization algorithm is proposed. The main idea of this method is to set unequal sensing period for each licensed channel, which has different usage pattern. We first construct the multi-channel states transition model by alternating renewal theory. Secondly, based on the continuous-time Markov chain, the choice of the multi-channel sensing period is modeled as a constrained multi-objective optimization problem. Finally, the multi-objective optimization problem is solved by genetic algorithms. The simulation results validate the performance of the derived objective function. When there are eight licensed channels in the target network, the proposed algorithm discovers 68.23% of free spectrum opportunities, which brings up to 17.68% more opportunities than the OFDM sensing method.