Fluorescence localization microscopy requires a huge number of image frames to ensure that different fluorescent particles are excited and emit light in different individual frames so that the fluorescent particles do not overlap in an individual frame, and the fine details of biological cells can be reconstructed by observing these fluorescent particles in all of the frames. This requires a long data acquisition time and huge storage for the collection of image frames. This paper proposes a novel method for estimating the coordinates of overlapping fluorescent particles so that the particle positions can be accurately determined, even when the density of particles in one frame increases. The proposed method involves the integration of model-based methods and exemplar-based method. The model-based methods estimate the particle positions accurately when the particle density is low and there are few overlapping particles, while the exemplar-based method estimates the particle number and positions accurately when the particle density is high and there is substantial overlapping. By integrating these two approaches, the proposed method can help save data acquisition time and storage. Experimental results show that the performance of particle position prediction also significantly improves when compared with model- based methods alone.