For　the　reconstruction　in　the　in-line　digital　holography,　there　are　three　terms　overlapping　with　each　other　on　the　image　plane,　named　the　zero　order　term,　the　real　image　and　the　twin　image　respectively.　The　unwanted　twin　image　degrades　the　real　image　seriously.　A　hybrid　phase　retrieval　algorithm　is　presented　to　address　this　problem,　which　combines　the　advantages　of　two　popular　phase　retrieval　algorithms.　One　is　the　improved　version　of　the　universal　iterative　algorithm　(UIA),　called　the　phase　flipping-based　UIA　(PFB-UIA).　The　key　point　of　this　algorithm　is　to　flip　the　phase　of　the　object　iteratively.　It　is　proved　that　the　PFB-UIA　is　able　to　find　the　support　of　the　complicated　object.　Another　one　is　the　Fienup　algorithm,　which　is　a　kind　of　well-developed　algorithm　and　uses　the　support　of　the　object　as　the　constraint　among　the　iteration　procedure.　Thus,　by　following　the　Fienup　algorithm　immediately　after　the　PFB-UIA,　it　is　possible　to　produce　the　amplitude　and　the　phase　distributions　of　the　object　with　high　fidelity.　The　primary　simulated　results　showed　that　the　proposed　algorithm　is　powerful　for　solving　the　twin　image　problem　in　the　in-line　digital　holography.