Region of interest (ROI) is commonly used in many application areas, including medical imaging, wafer defect detection, geographical information systems and computer vision and optical character recognition. Within a ROI may lie individual points of interest (POI). Many searching algorithms are proposed to find POI. Based on how they decide their trajectories, they can be categorized to sequential search and convergent iterative search. However, they are not efficient and fast enough. In this paper, we proposed a low cost DFS-based searching algorithm to speedup finding POI when the POI are not evenly-distributed. Representative algorithms of sequential search and convergent iterative search are implemented to compare with our approach. The results reveal that our algorithm can efficiently find the POI with less cost in diverse POI distribution models.