In　order　to　improve　visual　effect　of　single　night　color　image,　a　Retinex　enhancement　algorithm　based　on　weighted　least　squares　(WLS)　was　proposed.　Firstly,　this　algorithm　constructed　a　optimization　function　to　prepare　for　the　color　restoration.　Secondly,　it　processed　brightness　image　of　HSV　(hue,　saturation,　value)　color　space　using　the　WLS　optimization　framework　which　composed　of　brightness　and　gradient　characteristics　of　night　image.　Then,　the　scene　illumination　could　be　estimated,　and　reflection　image　could　be　obtained.　Finally,　it　finished　the　night　color　image　enhancement　after　the　color　restoration　and　compensation　process.　The　experimental　results　show　that　this　algorithm　has　a　good　ability　of　image　enhancement.　Compared　with　other　algorithms,　this　algorithm　also　has　the　characteristics　of　color　restoration,　Halo　effect　elimination,　noise　suppression　and　detail　preserving.