Abstract: Speckle noise is a common phenomenon in laser interferometry, which covers shape information of corresponding area of measured surface, resulting in measurement error. Aiming at characteristics of speckle noise in oblique laser interferometry, a recognition method of speckle noise based on object image is proposed. In this method, upper and lower thresholds of speckle noise are calculated automatically by statistic characteristics of gray scale distribution in effective measurement area and background area in object image. Position of speckle noise in interference fringe image is obtained based on mapping relation between object image and interference fringe image. A correlation experiment is designed to repair speckle noise region identified in interference fringe image, and eliminate phase transition between adjacent pixel points in a fringe period of wrapped phase diagram.