Getting (x,y) coordinates from a 2D image

[NZafr]

Hi,

I have an application that is based on Intel's Face Tracking sample of SDK 10.0.26.396.

My goal is to get (x,y) coordinate of a specific pixel in an image.

When I am in 3D mode, and apply QueryVertices() on sample->depth it is working well, and I get the pixel coordinate from vertices array.

However, when in 2D mode, although I have sample->depth, when I apply QueryVertices() on it I get a vertices array that is full of "garbage".

1. Why isn't the QueryVertices() working on the sample->depth image I am getting when I am in 2D mode?

2. Is there a way to get a valid vertices array from a color image?

Thank you,

Noam.

[idata]

Hello noamz,

Thanks for contacting Intel customer support. Please refer to page 21 of the SDK R2 release notes, specifically under the face detection, pose and Landmark tracking table. Here is stated that the 2D mode cannot be used. Here is the link to the release notes: http://registrationcenter-download.intel.com/akdlm/irc_nas/9078/release_notes_RealSense_SDK_2016_R2.pdf http://registrationcenter-download.intel.com/akdlm/irc_nas/9078/release_notes_RealSense_SDK_2016_R2.pdfThanks for the understanding. Best Regards, Juan N.

[NZafr]

Hello Juan,

Thank you for your quick response.

Indeed I see the comment you mentioned about working with face tracking in 2D mode.

However, this makes no sense because we are working with RealSense SR300 cameras for quite a long time, and have no problem doing face tracking in 2D mode.

Furthermore, I don't have any issues with face tracking in 2D mode, and my question was not about that.

Let me rephrase my question and ask it in a more general way... perhaps it will clarify what is my problem:

Given that I have the (x,y) coordinates of a pixel (that is image coordinates system), how can I get the real world coordinates (x,y,z) of that pixel (that is camera coordinates system)?

Thanks,

Noam.

[idata]

Hello Noam,

Thanks for your clarification and patience.To get the X,Y and Z coordinates please follow the information on the following Links:https://software.intel.com/sites/landingpage/realsense/camera-sdk/v1.1/documentation/html/index.html?doc_essential_coordinate_systems.html https://software.intel.com/sites/landingpage/realsense/camera-sdk/v1.1/documentation/html/index.html?doc_essential_coordinate_systems.htmlhttps://software.intel.com/sites/landingpage/realsense/camera-sdk/v1.1/documentation/html/index.html?member_functions_pxcprojection.html https://software.intel.com/sites/landingpage/realsense/camera-sdk/v1.1/documentation/html/index.html?member_functions_pxcprojection.htmlWith the commands on both you should be able to find the coordinates.Best Regards, Juan N.

[NZafr]

Hello Juan,

I am aware of the information referred by your links.

In fact, I am using some of the Projector functions for several purposes.

Again, the problem is that none of the Projector functions can take a pixel coordinate as an input and output the image coordinate that relates to it... or at least there is no Projector function that I know of.

Do you know if it is possible to convert a pixel coordinate to an image coordinate using RSSDK?

Thanks,

Noam.

[idata]

Hello Noam,

I have been checking your question with my development team. What you are trying to achieve is possible, there are notifications of customers able to do so. The Projection sample of the SDK has what you are looking for. Our best recommendation is to use it as a reference. Also here is the link to the projection sample in the user's manual. https://software.intel.com/sites/landingpage/realsense/camera-sdk/v1.1/documentation/html/index.html?pxcprojection.htmlBest Regards, Juan N.

[NZafr]

Hello Juan,

Thank you for making an effort to assist.

At the moment I am doing the conversion from pixel to real world coordinates using OpenCV functions, so I can do without the RSSDK functions.

However, I will review the link you've sent and try to understand how it can be done with RSSDK functions.

Thanks again,

Noam.