An inertially stabilized platform (ISP) is a gimbal used to hold a camera. It is essential for isolating the attitude sway during an imaging process. To make the platform more compact, the implementation of an ISP with a spherical mechanism has been proposed. However, this design introduces complex non-linear coupling characteristics, significantly increasing the complexity of control. This study aims to solve this problem and present an algorithm to achieve high-performance inertial stabilization control. To achieve this, kinematic and dynamic models were established, and the proposed control algorithm was then designed in three parts. Gravity compensation control was set up in the internal loop to counter the influence of unbalanced gravity moment. An adaptive sliding-mode-assisted disturbance observer (ASMADO) was also included in the joint space to decouple the influence of complex non-linear characteristics on the platform. Further, a feedback controller was added to the workspace based on the kinematic model. This design simplifies the control algorithm for novel ISPs with a spherical mechanism. It effectively compensates for the complex non-linear characteristics and enables superior inertial stabilization control. Experimental results show that the proposed method effectively decreases the motion isolation error for the line-of-sight of the camera compared to traditional control methods.