Particle streak velocimetry based on defocused imaging is proposed to measure three-dimensional (3D) velocity by recognizing defocusing parameter σ of straight and curve trajectories, called defocused particle streak velocimetry(DPSV). Assuming that σ varies linearly along the trajectory, it can be recognized by the surface fitting of the gray value distribution of particle images, and thereby the depth information of particles is estimated. Based on the linear fitting of straight trajectory, the arc fitting is proposed for the processing of images with curve trajectories, which are commonly captured in turbulent flow by PSV. The relevant image processing algorithm is developed and its correctness in the parameter recognition is validated by synthetic images. The results show that the relative error is about 9.4% in the presence of the noise from the experimental device in this paper. The linear relationship between σ and particle depth z is verified by experiments using an LED light and a diaphragm of 5 μm. Finally, the DPSV technique is applied to a jet flow field and its 3D velocity distribution is presented.