How can I add a floorplan image in rviz?

edit

I know I'm not answering your direct question, but couldn't you use the map_server (http://ros.org/wiki/map_server) node, remap /map to something else like /map_vis, and then use RViz to just visualize that map without interfering with anything else in your system?

I wrote a function using opencv in order to do so.

LUcorner is the position of the upper-left corner of the image in rviz. sizeSq is a scaling factor.

void sendImageToRviz(std::string frame_id, int id, ros::Publisher* marker_pub, cv::Mat src, double sizeSq, std::pair<double, double=""> LUcorner) {

visualization_msgs::Marker image;
image.header.frame_id = frame_id;
image.header.stamp = ros::Time::now();
image.ns = "image";
image.id = id;
image.action = visualization_msgs::Marker::ADD;
image.type = visualization_msgs::Marker::TRIANGLE_LIST;
image.scale.x = 1;
image.scale.y = 1;
image.scale.z = 1;

double pix = sizeSq / src.rows;
    geometry_msgs::Point p;
std_msgs::ColorRGBA crgb;

for(int r = 0; r < src.rows; ++r) {
    for(int c = 0; c < src.rows; ++c) {
      cv::Vec3b intensity = src.at<cv::Vec3b>(r, c);
      crgb.r = intensity.val[2] / 255.0;
      crgb.g = intensity.val[1] / 255.0;
      crgb.b = intensity.val[0] / 255.0;
      crgb.a = 1.0;

      p.z = 0;
      p.x = LUcorner.first + r * pix;
      p.y = LUcorner.second + c * pix;
      image.points.push_back(p);
      image.colors.push_back(crgb);
      p.x = LUcorner.first + (r + 1) * pix;
      p.y = LUcorner.second + c * pix;
      image.points.push_back(p);
      image.colors.push_back(crgb);
      p.x = LUcorner.first + r * pix;
      p.y = LUcorner.second + (c + 1) * pix;
      image.points.push_back(p);
      image.colors.push_back(crgb);
      p.x = LUcorner.first + (r + 1) * pix;
      p.y = LUcorner.second + c * pix;
      image.points.push_back(p);
      image.colors.push_back(crgb);
      p.x = LUcorner.first + (r + 1) * pix;
      p.y = LUcorner.second + (c + 1) * pix;
      image.points.push_back(p);
      image.colors.push_back(crgb);
      p.x = LUcorner.first + r * pix;
      p.y = LUcorner.second + (c + 1) * pix;
      image.points.push_back(p);
      image.colors.push_back(crgb);
    }
}
marker_pub->publish(image);

}

after trying various ways, i found that publishing the image as colored pointcloud is actually a very good approach.