This paper presents an unsupervised approach to accurately discover interesting places in a city from location based social sensing applications, a new sensing application paradigm that collects observations of physical world from Location-based Social Networks (LBSN). While there are a large amount of prior works on personalized Point of Interests (POI) recommendation systems, they used supervised learning approaches that did not work for users who have little or no historic (training) data. In this paper, we focused on an interesting place discovery problem where the goal is to accurately discover the interesting places in a city that average people may have strong interests to visit (e.g., parks, museums, historic sites, etc.) using unsupervised approaches. In particular, we develop a new Physical-Social-aware Interesting Place Discovery (PSIPD) scheme which jointly exploits the location's physical dependency and the visitor's social dependency to solve the interesting place discovery problem using an unsupervised approach. We compare our solution with state-of-the- art baselines using two real world data traces from LBSN. The results showed that our approach achieved significant performance improvements compared to all baselines in terms of both estimation accuracy and ranking performance.