To improve the signal noise ratio (SNR), it is important to enhance the intelligent analysis of MFL signal. If MFL data are processed directly, it usually has low precision due to the low and high frequency noise contained in MFL signal. It is discovered from the magnetic dipole model of infinite rectangular groove that there is a cross correlation between the original function and first derivative of the normal component and tangential component leakage magnetic field. Therefore, a MFL signal enhancement algorithm is proposed based on the cross correlation, which can enhance the signal near detect targets and suppress noise. Hence, the SNR of MFL signal is improved. The algorithm is validated and generalized by pipeline pull-through test data and in-service pipeline inspection data, respectively. Finally, some samples are collected from in-service pipeline inspection data and a signal quality estimation method is proposed which is suitable for evaluating the quality of the MFL signal. The enhancement algorithm is quantitatively evaluated by the samples and the estimation method. Experimental results show that almost all the samples are improved and most of the samples are improved by 10 dB at least.