Abstract: Linear octrees offer a volume representation of 3-Dobjects, which is quite compact and lends itself to traditional object processing operations. However, the linear octree structure for generating the representation of 3-D objects from three orthogonal views by using the volume intersection technique is dependent on viewpoints. The recognition achieved from matching object representations to model representations requires that the representalions of objects are independent on viewpoints. In order to obtain independent representations on viewpoint, the three principal axes of the object should be found frm the moment of inertia matrix by computing its eigenvectors. The linear octree is projected onto the image planes of the three principal views (along the principal axes) to obtain the three nomalized linear quadtrees. The object marching procedure has two phases: the first phase is to mtch the normalized ,linear quadtrees of the unknown object to a subset of models contained in .a library, utilizing a measure of symmetric difference; the second phase is to generate the normalized linear octrees of the object and these selected models and then match the normalized linear octree of the unknown object to the model with the minimum symmetric difference.