A fast algorithm based on the randomized Hough transform is presented in order to detect such dim target with parabolic track under the background of K-distributed sea clutter and thermal noise.This algorithm uses three randomly picked points and one searched discrete angle to extract parabola with arbitrary orientation

and employs the time information carried by the three data points to reduce useless samplings further

which speeds up the algorithm remarkably.A theoretic analysis is presented to analyze the detection performance of the algorithm.The results of performance analysis show the relations among the probability of detection

the probability of false alarm

the thresholds and the number of sampling.Simultaneously

the results also show that the detection performance of Hough transform will become the detection performance limit of randomized Hough transform

when the number of sampling increases to infinity gradually.