The boundary detection of continuous objects has become an important research challenge in Wireless Sensor Networks (WSNs), where improving the accuracy of boundary and reducing the energy consumption are the primary factors to be considered. To address this challenge, this article proposes a tow-stage boundary area detection scheme in duty-cycled WSNs, and sensor nodes are deployed in a dense fashion. Experimental evaluation result shows that the refinement procedure can refine the boundary area, where half of the initial boundary faces area should be reduced in most situations.