In this paper, the reassigned local polynomial periodogram along the frequency direction (RfLPP) is shown to be able to perfectly localize the chirp signal. Due to this property, the RfLPP is combined with the Hough transform for chirp signal detection in the parameter domain. Simulations of the RfLPP-Hough transform are given for chirp signal detection in white Gaussian noise and impulsive noise, with comparisons to the pseudo-Wigner-Hough transform (PWHT), the local polynomial periodogram (LPP)- Hough transform (LHT), and the reassigned LPP-Hough transform. It shows that the Rf- LPP-Hough transform can provide better performance with reduced computation time and is a proper candidate for chirp signal detection.