- From: Bryan B via GitHub <sysbot+gh@w3.org>
- Date: Wed, 16 Mar 2016 00:22:19 +0000
- To: public-media-capture@w3.org
bbernhar has just created a new issue for https://github.com/w3c/mediacapture-depth: == Accessing Camera Calibrations == I see the importance that we consider allowing the UA to access calibration data. It would enable many useful computer vision scenarios and brings the need to enable at-least three calibrations: * Spatial coordinate system. This gives us ability to project and unproject 2D image space to 3D camera space and map between locations of depth and the corresponding color image. Ideally, this wouldn't be done on a per-pixel basis and a coordinate system could represent the UA's world position too. * Lens distortion: The UA can remove distortion from the image as not all cameras use high-distortion lenses. * Camera re-sectioning: This can unproject 2D point to 3D space as ray (3D surface reconstruction from image or "skeletal tracking"). The `v1` spec does not mention how or which calibration is used unlike the more verbose SDK-specific APIs [1] [2] and it seems necessary to expose these intrinsic operations and properties that can be performed on streams that belong to the same group of 3D devices. @anssiko @huningxin WDYT? [1] https://msdn.microsoft.com/en-us/library/dn785530.aspx [2] https://software.intel.com/sites/landingpage/realsense/camera-sdk/v1.1/documentation/html/doc_essential_coordinates_mapping.html Please view or discuss this issue at https://github.com/w3c/mediacapture-depth/issues/110 using your GitHub account
Received on Wednesday, 16 March 2016 00:22:21 UTC