Objectives: To propose an algorithm for global optimized path planning for mobile robots on static and observable environments, based on image skeleton technique and using some other image processing algorithms. Methods/Analysis: The proposed geometric scheme is supported on the skeletonization of image of free space into a navigation environment, as a strategy to limit the number of possible paths from a starting to end points for a mobile robot. This approach, considerably simplifies the selection of appropriate paths, allows defining lines (edges) that can be utilized by graphs for the specification of paths. Several simulation and real test were done to prove the efficiency of the algorithm. For the real test, a prototype differential platform was build using a LEGO Mindstorms NXT robotics kit. Findings: The strategy demonstrated excellent performance and robustness over about 50 simulations and experimental tests, finding always a feasible and short path., but taking in average 6.4 seconds to find the solution. Novelty/Improvements: The proposed algorithm showed to work better than other tested algorithms like Voronoi and Bug, especially when the navigation environment is a maze. Keywords: Deep Search, Hybrid Scheme, Image Skeletonization, Navigation, Path Planning, Robotics