Photon　mapping　can　efficiently　render　high-quality　photorealistic　images　with　full　global　illumination.　It　stores　many　photons　in　photon　map　for　improving　estimated　qualities　of　complex　illumination,　which　leads　to　high　memory　overhead　and　long　computation　time.　The　usual　methods　of　illumination　estimation　of　photon　map　are　easy　to　produce　illumination　error　due　to　geometry　bias.　We　give　two　effective　solutions　for　these　existing　problems.　On　the　one　hand,　this　paper　presents　a　geometry-bias-based　reconstruction　of　photon　map.　Based　on　the　number　of　photons　and　the　curvature　variation　of　surface　in　every　non-empty　grid,　we　take　a　fraction　of　photons　in　grids　to　reconstruct　global　photon　map.　The　method　has　low　computation　cost　and　keeps　view-independence　of　photon　map.　Compared　with　the　non-reconstructed　photon　map　with　the　same　number　of　photons,　the　reconstructed　photon　map　has　far　better　effect　of　illumination　estimation,　and　can　effectively　reduce　all　kinds　of　geometrically　incurred　illumination　errors.　On　the　other,　we　propose　a　geometry-bias-based　illumination　estimation　of　photon　map.　Based　on　true　area　of　illumination　estimation,　which　is　intersection　area　between　bounding　sphere　of　photons　estimating　illumination　and　almost　coplanar　triangle　meshes　including　these　photons,　the　estimated　illumination　effect　can　be　largely　improved　by　reduction　of　geometry　bias　of　illumination　estimation.