A novel method is developed to improve the accuracy of micro-resolution particle image velocimetry (PIV) in microfluidics measurements. This method utilizes the Laplacian of Gaussian method and image-processing techniques to eliminate the background scattering noise. A high signal-to-noise ratio image has been obtained. This technique is especially suitable for improving micro-resolution PIV in micro, two-, or multiphase flow conditions, such as for submicron bubbly flow measurements in a microchannel. The method can easily be implemented with minimal modification of the conventional PlV system. The results of simulation and experiments demonstrated the feasibility of this, to our knowledge, new method.