This paper presents a stabilization framework integrated with the estimation of the terrain inclination to balance a humanoid on the changing slope as an extension to our previous study. In this paper, the estimation of the terrain inclination is improved for walking in place on an inclination-varying slope. A passivity based admittance control utilizes the force/torque sensing in feet to actively regulate the impedance at the center of mass to stabilize the robot. The logic-based inclination estimation algorithm uses the feet to probe the terrain and deals with the under-actuation. The equilibrium set-point in the admittance control is regulated based on the detected inclination. The effectiveness of the control framework is validated on the humanoid robot COMAN and demonstrated by estimating the terrain inclination, coping with the under-actuation phase, adapting to the slope with changing inclination during both standing and walking. Experimental data are analyzed and discussed, and the future work is suggested.