Abstract: The parabola object feature detection of optical image is a difficult research topic in computer vision fields. A parabola feature detection algorithm based on the least square method is presented. Using Canny algorithm to detect edge, then the least square method is used to compute the coefficients of the parabola equation. The conic is constrained by parabola and the algorithm, so it can get the optimal solution without iteration computation. Finally the coordinate transformation of the parabola equation is computed and the parameters of parabola can be achieved. The experimental results of emulation data and real image show that our algorithm can efficiently detect the parabola object feature and compute the feature parameters in the optical image, and the calculation accuracy of each parameter is above 98%.