Most of algorithms of lane detection mainly aim at the scene of daytime. However, the camera is very sensitive to the light which causes such algorithms are unstable for the lane detection at night. This paper proposed a lane detection algorithm that largely improves the lane detection’s performance when it used at night. Step one is image processing. According to the feature of lane’s color, we focus on the image’s white and yellow parts and transform the original image to grayscale image. When the image’s value of grayscale is relatively small, an image enhancement method will be used to improve the contrast between lane line and road. Then, extracting a region of interesting from image (ROI) according the fixed location of camera mounted inside the car. In ROI, we used a creative algorithm named Correlation filter to remove image noise and remain the feature of lane. In step one, the filter matrix looks like “[0 1 1;-1 0 1;-1 -1 0]”. Final, extracting the line by the Hough transform and acquiring the equation of lane by fitting spots obtained from Hough. Step two, after second frame of image, we used the Kalman filter to trace the detected lane, which improving the efficiency and the accuracy of image processing. In step two, the filter matrix’s dimension will change according with the equation’s parameter which acquired from last frame. Finally, we test this algorithm in a real vehicle and get an excited result. After second frame of image, this algorithm’s image processing speed approximately keeps on 15 frames per second and average accuracy reaches at 93.52%.