Automatic target detection in SAR image is the first stage in the ATR (Automatic Target Recognition) system. It is very important to improve the computing efficiency of the whole system. A conventional automatic target detection method is CFAR (constant false alarm rate) which is based only on the amplitude, discards the phase and thereby sacrifices half information of the SAR imagery. Almost each CFAR brings a large number of false alarms. 2L-IHP (two-look internal Hermitian product) is a new method which exploits the phase and amplitude information. In this paper, both CFAR and 2L-IHP algotithms are compared with an emphasis on 2L-IHP. The 2L-IHP algorithm for target detection is discussed briefly first. In order to apply it widely, it is modified to circumvent the constraint of the data. For polarimetric SAR images, PWF (Polarimetric Whitening Filter) is used to generate an image for target detection. According to ADTS data set, these algorithms are analyzed and evaluated respectively. C-、L- and P- band AIRSAR data are also processed, and some tentative conclusions are derived: Considering the penetrating ability of the microwave, L-band got the best detection accuracy of the three bands; HH polarization performed better for target detection than VV and HV polarization; Moreover, because more information is contained in polarimetric SAR data, here, we conclude that polarimetric SAR data have advantages over single polarized SAR data when they are applied for target detection. But all these conclusions need further validation.