With the continuous development of urban traffic and the improvement of intelligent transportation systems, the accurate extraction and analysis of road traffic markings becomes more and more important. At present, road traffic marking is extracted by collecting dense point cloud data with highthread LiDAR. However, due to the high cost of the high-thread LiDAR and the large amount of point cloud data, the production efficiency of high-precision maps is low, and updating is complex and difficult. Therefore, a traffic marking extraction algorithm based on image and point cloud data fusion is proposed. The innovation of this study lies in the introduction of image processing for traffic marking edge detection, and the adaptive threshold segmentation of point cloud data using the maximum inter class variance method. The algorithm first uses an edge detection algorithm on the ground of the Canny operator to extract edge information of traffic markings from road images. Meanwhile, it uses the maximum inter class variance method to extract the internal point coordinate information of traffic markings from point cloud data. Subsequently, clustering segmentation, geometric correction and other techniques are used to process the fused data, thereby achieving automated extraction of road traffic markings. The results show that on the TuSimple dataset, the false detection rate and missed detection rate of the edge detection algorithm on the ground of the Canny operator are only 3.84% and 2.79%, respectively. Meanwhile, in the TuSimple dataset, the peak signal-to-noise ratio and structural similarity of the maximum inter class variance method were as high as 39.25% and 9.23%, respectively. In addition, the average time it takes for the algorithm proposed by the research institute to extract the edge points of a single road traffic marking is only 5.69ms. Especially, the proposed method achieved a maximum improvement of 14.07% in the accuracy of traffic marking edge detection. The method proposed by the research institute effectively reduces false positives and missed detections, and can accurately and efficiently extract the edges of road traffic markings, providing a new and reliable solution for the automated extraction of road traffic markings.