Being the preliminary task of holographic image processing, various approaches for the holographic image analysis have been proposed. This presentation describes an on—line image processing system for the automatic distribution analysis of particles in particle holograms. In this system, the particle holograms are obtained using in—line retrieval method. To find the approximate locations of particles, or seed points, the holographic images are first processed by the Sobel operator, and automatically thresholded. Then, using the seeds and a bi—threshold region growing technique, the edges and the preliminary parameters of the candidate particles are obtained by edge tracking in the original images. To remove the speckle noises caused by in—line retrieval method, the variance of the radius of the candidates and the texture of the candidate areas in the Sobeled images are analyzed. After that, the radial intensity profile of the candidates and the clearness of candidate neighboring areas in the original images are analyzed to remove particles out—of—focus. At last, the candidates are verified, and the statistical data of the verified particles are calculated. Experiment results are given to show the efficiency of the approach described in this presentation. Key Words: Particle Hologram, In—Line Retrieval Method, Segmentation, Region Growing.