Abstract: To satisfy the personalized requirement of privacy preservation, an algorithm for preserving trajectory privacy is proposed, which is based on the technique of Canonical Correlation Analysis (CCA). In this algorithm, the trajectories, which are considered to be insensitive by the data producer, are published directly. The protection of privacy is only imposed on the trajectories, which are considered to be sensitive by the data producer. To this end, a latent variable is first obtained from both the insensitive and sensitive trajectories. Then based on the latent variable, a lot of trajectories are generated to replace the sensitive trajectories. The advantage of this algorithm is that it can not only respect the privacy preserving wish of the data producer, but also obtain high quality trajectories. Experiments on real-world data set of trajectories were carried out, and the results demonstrate the effectiveness of the proposed algorithm.