The Remote Stereo Viewer (RSV) is a client/server application that provides immediate access to large sets of high-resolution stereo images using a QTVR style interface. It assumes that the user will view only a “small” selection of the images and consequently does not warrant downloading the entire image set.
Capabilities & Requirements
Image sets are stored on the server as JPEG image files. The images are logically arrainged into a multi-dimensional grid. Every image set has at least one dimension consisting of at a full revolution (typically at 5 degree increments) around an object. Consecutive images along this axis are used to construct the stereo view of an anatomical object. Extra dimensions may contain other views of the object such as layers of dissection or magnified views.

The user interface maps left/right motion of the mouse onto the rotation dimension. Vertical motion of the mouse can be selectively mapped onto other dimensions of the data set. The system also includes a label server that provides structure identification tags.

RSV clients are able to form collaborative groups so that one user can assume the role of leader for the group. When active, group clients follow the leader through the image set. In addition, the leader controls a “pointer” which can be used to mark locations in the image. Clients are free to enter and leave this mode as required.

The various components of the Remote Stereo Viewer are incorporated into the Information Channels framework.

The client maintains both a tcp and udp connection to the server. Image requests are sent over the tcp connection. Images are transported over the udp connection. Group collaboration is implemented by exchanging state information between clients over a multicast address.

Currently, image transport occurs at approximately 35Mbps, which corresponds to the time required to decompress the high-resolution images. Group collaboration information consists of approximately 50 byte packets sent by the current leader 15-20 times a second. Network loss and latency directly impact the rate at which images can be acquired and displayed by the client.

This animation shows the two dimensional grid of hand images arranged into a cylinder. The animation shows a typically path through the image set, starting with the undissected initial image, moving two layers into the dissection, rotating the hand, moving deeper into the dissection and rotating back.
Publications & Presentations

Last Modified: 6-Apr-03, Server Contact:
Copyright © 1995-2003, The University of Wisconsin - La Crosse Scientific Visualization Project