Effective retrieval of 3D objects is an important difficult problem. In addition to considering object-global similarity notions, partial similarity among 3D objects has been addressed in recent research. In this paper, we introduce a simple, yet effective retrieval system design based on extending standard global 3D descriptors with local descriptions of the objects. The approach is mapped to a novel retrieval interface concept that supports both global and local search for 3D objects. Also, an implementation of the concept is presented. This design performed based on image connectivity. Multiple views of an objects are taken as set of input data. Initially while processing two images, the common pixels between both images are obtained. Then the common pixels are merged into single pixel. That means the two images forms a combine image. That gives the enlarged view of the Object. That makes the view in enlarged dimension. Obtained image gives the dimensional valued image performed based on SURF features. And also the image frame realigned in size depends on matched pixels. For this matching of common pixels, SURF and RANSAC algorithms are used. And then we can combine random images with obtained output.