Since a multi-view lens-plate 3D image is displayed with picture data of many views, it is necessary to compress the data before transmission and recording. In this paper, a method of data compression using the geometrical relations between views of still, opaque, and color objects is proposed. By computer simulation, the estimated number of bits or a compressed data is about 1.5 times greater than that of a stereoscopic image under the condition of the same resolution. Experimental data obtained by camera were compressed and expanded by this method, and 3D feature of the object was reproduced.