A server (1) analyses the request information transmitted from a client (11) by means of a request analyzer 4 and selects the necessary image data from multiple viewpoint image data (2) to output it to an image generator (3), where image data for the requested viewpoint is generated by interpolation to be output to an image synthesizer (5). In the image synthesizer (5), a plurality of images data are synthesized in a form suitable for encoding. An encoder (6) encodes the image data at a suitable bit rate and transfers it to a network (7). The client (11) receives the encoded image data, decodes the data through a decoder (12) and outputs the decoded image data to an image processor (13), where the image data is converted into an appropriate form in conformity with a stereoscopic display format and is displayed on a display unit (14). The client (11) also includes an input unit (16) for change of the viewpoint, and transmits the request information of viewpoint alternation to the network (7) by means of a request output unit (15). Thus this system enables observation of a stereoscopic image viewed from an arbitrary viewpoint even with a mobile terminal or the like.