Efficient and accurate geometric inspection planning is a critical challenge in robotic automation, particularly for quality assurance and metrology applications. This paper presents a markerless 3D inspection planning framework that addresses the Viewpoint Planning Problem (VPP) using a greedy optimization algorithm. The proposed method discretizes the object and viewpoint search space into point clouds, transforming the VPP into a Set Covering Problem (SCP). A novel registration stability constraint ensures robust markerless alignment of scans, eliminating the need for time-consuming marker placement. Markerless registration is enabled by the smart selection of viewpoints, which ensures optimal coverage while maintaining registration accuracy. Validation of the framework demonstrates its effectiveness in generating near-optimal inspection plans and achieving reliable, high-quality reconstruction without physical markers.