The object segmentation of defocused images with projected illumination patterns is a primary component of 3D object recovery. A segmentation algorithm

which combines multi resolution analysis and fuzzy clustering

is proposed based on features extracted from defocused images with illumination patterns to increase the segmentation accuracy. This approach is carried out through two stages: firstly building the relational matrix of multi resolution images on their similarity measures and then implementing FCM classification at lower resolution level. The proposed technique provides a novel way of avoiding the difficulties of direct segmentation. And it increases the accuracy of object segmentation. It is shown in experiments that the proposed algorithm is of effectiveness.