The invention discloses a detection method for surface defects of an engine cam shaft part. The detection method comprises a detecting algorithm for ring surface defects of upper and lower end surfaces, an algorithm for detecting whether a thread exists at the bottom or not, and an algorithm for detecting whether scrap iron is remained in a bottom hole or not, wherein the detecting algorithm for the ring surface defects of the upper and lower end surfaces adopts a possibility Hough conversion method combining with gradient information to carry out circle center positioning on the upper and lower end surfaces; central circular holes are removed; then images of the ring surface of the upper and lower end surfaces are pre-treated by adopting homomorphic filtering and image noises are removed by using opening operation; defect regions are clustered and defect types are judged. According to the detection method, a coaxial light source, an annular light source, a point light source and an area-array camera are used for carrying out polishing and image acquisition on the end surface of the part and the bottom hole. A dividing device and a photoelectric switch are used for controlling an intermittent rotary mechanical structure of a circular disc, so that the automatic detection, with the resolution rate better than 0.1mm and the single piece detection time less than three seconds, on surface cracks, scratches and spots on the surface of a workpiece, and on whether the scrap iron is remained in the thread of the bottom hole of the part and in the hole or not can be realized.