How to zero out pixels in RGB image based on the corresponding depth values in a faster way ? RRS feed

  • Question

  • Hi,all

    A part of my application is to zero out the RGB image pixels which are beyond a certain distance range using kinect V2 SDK, the official CoordinateMappingBasics demo gave me some hints to do so, but in fact use the coordinate mapper to map the cameraspacepoint to a depthspacepoint and zero out the corresponding pixels in a RGB image which has resolution of 1920*1080 are time consuming, even the official demo can not get silky smooth video performance(roughly 20~25fps).

    Since I have other time-consuming operations to do after the "zero out" thing, so I am looking for faster way to do that:

    PLAN A is to use more efficient code to deal with that, I am new to C#, does process each row in parallel can have a great enhancement ?if it does, how ?

    PLAN B is to resize the RGB image to a smaller scale resolution, let's say 640*480. But I doubt that MapColorFrameToDepthSpace method can still work after I resized the image.

    Is there any suggestions?Thanks in advance .

    • Edited by Jialiang M Monday, October 19, 2015 1:23 PM
    Sunday, October 11, 2015 6:48 AM

All replies