A handheld multi-sensor system, including GPR, is an effective solution for landmine detection. But it is difficult to show the visualization images because of the irregular measurement data acquired by human being operator. To deal with the problem, an interpolation is a common choice to create grid data set. But generally the common interpolation algorithms can not offer the good signal-clutter ratio in a complicated situation, although it can offer grid data set for visualization. Also the common 3-D interpolation algorithms consume a lot of processing time. Here we propose a 2.5-D interpolation algorithm that uses the Kirchhoff migration integral to produce a nonlinear interpolating polynomial. Comparing with common linear interpolation and cubic interpolation, the new algorithm can achieve the better results. Also the algorithm is faster than common 3-D interpolation algorithm. Lastly, the algorithm was applied to the real measurement data set and got high quality images.