A novel approach is proposed to detect infrared dim target from cluttered background by using morphological band-pass filtering and scale space theory. The infrared image is pre-processed by means of morphological band-pass filter, which results in regions of interest (RoI) containing dim small targets. Then, difference-of-Gaussian function is adopted to obtain scale space of pre-processed infrared image. Scale space maximum detection is then performed to generate candidate targets with their positions and scales. Infrared dim small target detection is achieved by using thresholding signal-to-clutter ratio of candidate targets. Experimental results on real-world infrared images and comparisons with state-of-the-art methods can demonstrate the effectiveness and robustness of the proposed approach.