The invention discloses a vehicle violation line pressing detection method based on computer vision, and the method comprises the steps: 1, obtaining an image, carrying out the filtering and equalization processing, weakening the noise influence, and enabling the gray level distribution of the image to be uniform; 2, carrying out edge detection on the image processed in the step 1 by adopting a Canny edge detection algorithm, and carrying out image segmentation by adopting a threshold value; 3, using the Gaussian mixture background model to perform background modeling including background learning, background pixel matching, discontinuous contour processing and foreground cavity filling; 4, using Hough transformation to obtain linear information in the model obtained in the step 3, and performing AND operation on the linear image and a contour image detected by an original image Canny operator; 5, extracting features by using the moments of the image, and screening out a target straight line according to the length and angle of a straight line to be obtained during lane line detection by using the moment characteristics; and 6, judging whether line pressing violates rules or not according to the shielding condition of the lane solid line and the contour characteristics of the line pressing object.