Abstract: In terms of the features of flash X-ray radiographs, the typical edge detection algorithm in digital image processing can only provide thick edges and is noise susceptibile. Thus an edge detection method based on Facet model for detonation experiment images is proposed. This method absorbs the merits of its congeneric method and introduces orthogonal basis to simplify operation. It uses surface fitting methods to approximate locally the image intensity function, and then makes decisions regarding the edge locations using the partial derivatives of the fitted surface. The result shows that the obtained edges of the image are thin, smooth and continuous. The algorithm has a better immunity to the noise jamming of images, and can give accurate localization for isolated edges.