Image understanding is the science of automated extraction of information from images. This science includes three aspects: rudimentatory processing, intermediate processing and advanced processing. This paper discusses the problem about how to trace the contour of bilevel-images, which belongs to rudimentatory processing. Contour tracing is to extract the outside-edges of the choosed object region. Then, some other methods used for contour tracing are presented and a new general tracing contour algorithm quite different from the existing methods is proposed. Good experimental results have been obtained. My original intention is to be innovation.