How to detect infrared dim targets accurately under complex background and low SCR condition is of great significance for the development of precision guided weapons and infrared warning. In order to improve the SCR and detect the dim targets effectively, a new method for infrared dim target detection based on the gray contrast between the central region and its neighborhood was proposed. The contrast of the target was improved by calculating the contrast map and saliency map of the input image while suppressing the background clutter. The adaptive threshold was set on this basis to separate the dim targets. Experimental results show that the proposed method can achieve higher detection rate and lower false alarm rate compared with conventional LCM(Local Contrast Measure) method. The proposed method has an outperformance compared with other algorithms, especially in the case of complex background.