Aiming at the problem of missing point cloud data when measuring highly reflective surfaces with traditional sinusoidal stripe structured light, this paper proposes a 3D reconstruction method of highly reflective surfaces based on binary encoded stripes. Firstly, half-periodic sinusoidal stripes are used to replace the traditional sinusoidal stripes and binary coding is applied to them, and only binary stripes need to be projected during the whole measurement process, which avoids the complex nonlinear correction process and enhances the system's anti-interference ability. Secondly, the number of exposures is reduced by the introduction of the quantized exposure time selection strategy, which shortens the measurement time. Lastly, the high-dynamic-range images are acquired by combining with image mask fusion technology. The experimental results show that the number of complementary point cloud voids is 20853 and the growth rate of the point cloud is 8.45% using the method of this paper for the reconstruction of highly reflective surfaces of ceramic cup, and the proposed method has a certain degree of universality for different material objects.