Because of noise and clutter, the infrared target detection even becomes more difficult. In this paper, we present an
automatic seed selection method based on an improved mountain cluster algorithm to be employed in infrared target
segmentation. The original image was projected to x and y coordination firstly. Then the modified mountain clustering
algorithm is employed on the projection data to obtain the cluster center which is applied as the seed point for region
growing. This approach has been performed on a field infrared image. The experimental results show that the proposed
method can provide satisfactory segmentation result.