Target tracking with a mobile robot has numerous significant applications in both civilian and military. Practical challenges such as limited field-of-view, obstacle occlusion, and system uncertainty may all adversely affect tracking performance, yet few existing works can simultaneously tackle these limitations. To bridge the gap, we introduce the concept of belief-space probability of detection (BPOD) to measure the predictive visibility of the target under stochastic robot and target states. An Extended Kalman Filter variant incorporating BPOD is developed to predict target belief state under uncertain visibility within the planning horizon. Furthermore, we propose a computationally efficient algorithm to uniformly calculate both BPOD and the chance-constrained collision risk by utilizing linearized signed distance function (SDF), and then design a two-stage strategy for lightweight calculation of SDF in sequential convex programming. Building upon these treatments, we develop a real-time, non-myopic trajectory planner for visibility-aware and safe target tracking in the presence of system uncertainty. The effectiveness of the proposed approach is verified by both simulations and real-world experiments.