Terraced fields are a kind of soil and water conservation measures explored by humans on sloping fields. The construction of terraces largely develops the agricultural growth potential of sloping arable land, which has the functions of water storage and soil conservation. Due to the difficulty in obtaining information such as the number of terraces and distribution of area, it is difficult to carry out the quantitative research on the terraced fields. With the continuous development of unmanned aerial vehicle (UAV) technology, it becomes possible to access high-precision terrain information. Based on the UAV orthorectified images and slope data calculated by digital elevation model (DEM), the rough contour of terraced fields was extracted by Canny edge detection operator, and the false edges of terraced fields were removed according to the structural characteristics of terraced fields. According to edge strength superposition and edge connection operation, the terraces were divided by region growing algorithm. The method effectively solved the problems of uneven terraced fields in the hilly areas, interference of the surface sediments and complicated spectral characteristics of the images. Compared with the field data of terraced plots marked by hand, the results showed that the total accuracy of the proposed algorithm in terraced fields can reach 84.9%. The research result can provide a solution for the rapid mapping of terraced fields.