Localization technique is required to obtain the unknown nodes position in wireless sensor networks. Linear Intersection and Centroid are two selected methods on estimating the unknown position from its corresponding anchors. The execution time of these methods is acceptable. However, the error of unknown nodes position yields worst performance. We exploit likelihood estimator in refinement process of linear intersection method to obtain the sufficient error. Uncertainty impact on centroid method will be minimized by inserting the weighted factor on it. Our proposed method results in 50% reduction of average position error in grid deployment scheme, although larger execution time is needed. This paper provides an efficient method to solve the localization in small-scale deployment.