The quality of the road surface is one of the most popular problems around the world, since the life of a modern person is almost impossible to image without transport. There are many technologies to solve this problem. However, it works on the basis of various algorithms, which include contour detection methods. Each algorithm has its own characteristics. One of the most important characteristics of these methods is the accuracy of pattern recognition (classes) in the image. To have an approximate idea of how accurate a particular algorithm is, it is necessary to conduct a comparative analysis. The following methods is chosen for the study: Canny Operator, Kirsch Operator, Marr-Hildreth algorithm, Prewitt Operator, Sobel Operator. A total of 140 images are used for the experiment (all photos are different, stored in the corresponding folders on the hard disk, each size is 200x100/20000 pixels). 4 categories of images of the road surface are used: without damage, with cracks, with potholes, with ruttings. 35 photos are allocated for each type. For each category, in turn, it is divided into two groups: template (10 images, comparison is carried out with it) and test (25 images). The stages of the experiment are described in more detail in the section "Methodology". As a comparative characteristic, the indicator of correct answers for each of the selected contour selection algorithms is selected. The Kirsch Operator has the best value (46.2%), and the Prewitt Operator has the lowest value (38.2%). However, the average percentage of correct answers for each of the methods is quite small - less than 50%. The following question appears: "Is it possible to develop (or already exists) a method that processes the image once and at the same time has an average accuracy rate other than 50%." The answer to it can be obtained during further research.