As a promising technology to implement Dynamic Spectrum Access (DSA), Cognitive Radio (CR) is envisioned to evolve to be Green. In this paper, the Cooperative Sensing Scheduling (CSS) problem for Cognitive Radio Network (CRN) is analyzed by taking both the performance and energy consumption of spectrum sensing into consideration. We model the CSS problem under a practical scenario where primary user (PU) channels are heterogeneous in terms of channel protection criteria, idling probabilities and channel capacities. With the objective to find an optimal and efficient secondary users (SUs) assignment scheme, we formulate the CSS problem into a nonlinear integer programming problem. The intricate two-dimensional trade-off among energy efficiency, sensing performance and spectrum opportunity exploration is presented and properly tackled. We explore the inherent property of this problem by adopting a two-step approach. In the first step, by fixing the number of SUs participating in sensing, the problem is converted to an M-concave problem and an efficient algorithm for optimal SU assignment is proposed. With this scheme, we then investigate the appropriate number of SUs participating in sensing in the second step. Based on these two steps, the final SU assignment scheme is proposed. The optimality of this assignment scheme is proved theoretically and verified numerically.