When machine vision technology is used to detect the surface defects of parts with omnidirectional curved surface characteristics such as ‘O’ sealing rings, it shows a poor consistency of surface imaging. To achieve the full-automatic and reliable detection of surface defects of ‘O’ rings used in aerospace systems, the curved surface imaging theory and the surface defect detecting methods of ‘O’ rings were investigated. Firstly, the surface of ‘O’ ring was divided into small pieces like small grids. Based on the bidirectional reflectance distribution function of surface pieces, the radiation strength calculation formula of surface pieces was derived under the condition of point light irradiation. Then, the radiation strength calculation method of surface pieces suitable for the space continuous light irradiation was obtained through a hemisphere integral. Finally, combining with the ‘O’ ring surface equation, the imaging theory of ‘O’ ring surface was established and the imaging and detecting methods for ‘O’ ring surface defects based on three lens & camera units were proposed. The imaging experiment results of the segmental arc surfaces of ‘O’ rings indicate that the theoretical brightness distribution is consistent with the actual brightness. The average brightness error corresponding to the vertical lens & camera unit is 6.8 and its standard deviation is 12.6. The maximum average brightness error corresponding to the gradient lens & camera units is 19.4 and its standard deviation is 10.3. The defect detecting experiment results show that the proposed method detects reliably the defects in any location of the surface of ‘O’ rings and the proposed detecting system implements the automatic detection of the sealing rings in aerospace systems.