Aiming at the difficulty of multi-scale detection of ship targets in infrared polarization images under complex interference conditions of sea-sky scene, a ship target detection method was proposed based on guided filtering and adaptive scale local contrast method. First, by using the intensity information as the guiding information to guide and filter the infrared polarization image obtained the fusion image with higher local contrast and local signal-to-clutter ratio. Then, based on the vertical gradient characteristics of the fusion image, detecting the sea-sky line method was proposed so that sea-sky line weighting suppressed the sea clutter on the fused image. Finally, based on the single-scale local contrast method and the ship target proportional feature, the adaptive scale local contrast method was proposed. When the scale matched the target, the response was the largest, and the maximum scale was determined by the response of the target at different scales. The experimental results show that the local contrast and local signal-to-noise ratio of the image are improved by the guided filtering fusion method. Compared with the typical detection methods, our method can effectively suppress interference and detect different scales of ship targets in sea-sky scene with high robustness and accuracy. The detection accuracy ratio and false alarm ratio are 95.0% and 3.5%, respectively, which provides a new method for infrared polarization image target detection.