According　to　analyze　vision　system　of　hazard　cameras　for　lunar　rover　BH2　and　characteristics　of　working　environment,　firstly,　intrinsic　and　extrinsic　parameters　of　camera　are　accurate　calibrated　and　image　distortion　is　corrected　in　this　paper;　Secondly,　images　are　processed　according　to　Bayer　filter,　image　rectification,　LoG　filter,　pyramid　delaminating,　photometric　consistency　dense　matching;　Finally,　the　paper　can　obtain　high　precision　and　dense　disparity　map.　Furthermore,　the　paper　uses　three　sets　of　experiment　results　on　obstacle　avoidance　cameras　vision　system　of　lunar　rover　BH2　to　validate　the　scheme.　the　experimental　results　show　that　the　system　possesses　the　features　with　less　calculation　amount,　high　reliability,　robustness　of　illumination　change　and　environmental　noise,　which　can　quickly　and　reliably　achieve　dense　disparity　map　and　meet　the　real-time　avoid　obstacle　requirement　of　BH2.　©　2009　Springer-Verlag　Berlin　Heidelberg.