image_proc not working as expected

edit

Hello everyone,

I'm trying to use image_proc to rectify my image, coming from a folder.

Here is my launch file:

<launch>

<group ns="calibration">
    <rosparam command="load" file="/home/george/GOPRO/chessboard/calibrationdata/cal.yml"/>
    <!-- <node pkg="green_eye" name="camera_info_pub" type="camera_info_pub" output="screen" /> -->
</group>

<node pkg="green_eye" name="file_monitor" type="file_monitor" output="screen"/>
<node pkg="green_eye" name="image_processor" type="image_processor" output="screen"/>
<group ns="camera">
    <node pkg="image_proc" name="image_proc" type="image_proc" output="screen"/>
    <node pkg="image_view" name="image_raw" type="image_view" output="screen">
        <remap from="image" to="image_raw" />
    </node>
    <node pkg="image_view" name="image_mono" type="image_view" output="screen">
        <remap from="image" to="image_mono" />
    </node>
    <node pkg="image_view" name="image_rect" type="image_view" output="screen">
        <remap from="image" to="image_rect" />
    </node>
    </group>
</launch>

The file_monitor node checks for new images in a folder and publishes the image path. image_processor reads that path and uses openCV to read the image and publish it to the /camera/image_raw node. It also stamps and publishes the CameraInfo topic in the /camera/camera_info path. From there on, I expect image_proc to do its job, but it fails.

While the image_raw and image_mono topics are visible by the image_view, image_rect always stays grey (in fact it is never published).

I also get the following warning (not actually pasted exactly)

  [image_transport] Topics '/camera/image_mono' and '/camera/camera_info' do not appear to be synchronized. In the last 10s:
  Image messages received: 4
  CameraInfo messages received: 4
  Synchronized pairs: 0

image_mono seems to be a product of image_proc, but not actually a topic I publish myself, which lags image_raw by some 10s of ms and is displayed correctly. However, image_raw should be the only input of image_proc that I have to stamp. Thus, I don't know why this warning should appear, or whether it affects my result.

Any ideas welcome, Thanks in advance!

Edit -------------------------------------------------------------------------------------------------------------------------

I managed to have the intended result with the cv::undistort API, but the solution is less modular and I'm not sure whether the result is as sound as the well-documented image_proc, so I'll leave this open.

Edit 2 ----------------------------------------------------------------------------------------------------------------------

Here is the code which I use to fill the camera_info message, having the converted .yaml file loaded in the parameter server:

pub_img_raw = it.advertise("/camera/image_raw",1);
pub_cfg = n.advertise<sensor_msgs::CameraInfo>("/camera/camera_info",1);

sensor_msgs::CameraInfo cfg;
XmlRpc::XmlRpcValue list;
int i;

cfg.header.frame_id = "camera";
double temp;
ros::param::get("/calibration/image_height", temp);
cfg.height = (uint)temp;
ros::param::get("/calibration/image_width", temp);
cfg.width = (uint)temp;
ros::param::get("/calibration/distortion_model", cfg.distortion_model);
ros::param::get("/calibration/distortion_coefficients/data", list);
cfg.D.clear();
for (i=0;i<5;i++) {
    cfg.D.push_back((double)list[i]);
}
ros::param::get("/calibration/camera_matrix/data", list);
for (i=0;i<9;i++) {
    cfg.K[i] = list[i];
}
ros::param::get("/calibration/rectification_matrix/data", list);
for (i=0;i<9;i++) {
    cfg.R[i] = list[i];
}
ros::param::get("/calibration/projection_matrix/data", list);
for (i=0;i<12;i++) {
    cfg.P[i] = list ...