i have problem when color align with depth

[ylee39]

when I use the function CreateColorImageMappedToDepth(),object was overlapping partially in the color image ,easpcially when i was closed to realsense sr300 or object border

[MartyG]

I will link RealSense stream programming expert jb455 into this discussion to seek his input, since he can give a better answer on this subject than I can.

[jb455]

It's hard to tell from your picture - is your problem similar to /thread/111925 this? If so, it's likely to be an issue with the factory calibration, which as far as I'm aware Intel still hasn't done anything to fix.

Also note that the minimum distance the SR300 is rated to is 200mm and you often get weird stuff in the depth image when closer than that.

[ylee39]

Thanks ,My question is when aligned the color with the depth ,depth data is not stable, may be in bright area may also be in the overlapping area.so the distance of two points is wrong. A large area or big object will not have this kind of situation, because there is no shadow,the aligned picture is not overlapping.the reason is shadow?How can I get stably corresponding depth data ?

[idata]

Hi Cindyisme,

Have you tried using the Projection sample? It is located under RSSDK\sample\core directory, here you can find more information: https://software.intel.com/sites/landingpage/realsense/camera-sdk/v2016r3/documentation/html/index.html?sample_projection_cpp.html Sample: Projection [C++]. Additionally, please consider that the camera performance will vary depending on certain conditions such as ambient lighting conditions, the object's IR reflectivity and texture, backgrounds, distances. Regards,-Yermi A.

[ylee39]

Hello Intel Corporation，

I have tried using the Projection sample and librealsense,I found the same thing when aligned color with depth.According to a special depth point and UVMap, I could find the same color coordinates (i,j)in the UVMap (except (-1,- 1)).UVMap is one to many.So I find that why we can see the overlapping region .

[idata]

Hi Cindyisme,

I was able to reproduce this behavior with some objects. The quality of the color to depth mapping is affected by the object dimensions and material. For example, thin plastic objects show that double image phenomenon while other objects of similar size but different material, like wood, do not show it. Proximity to the camera also affected the result--the closer the object, the more duality we observed. Holding the object at 50cm or further from the camera yielded better results. It is a limitation of the technology in the SR300. We tried the same experiments with the R200 camera and it performed better. Regards,-Yermi A.

[ylee39]

Thank you for your reply,I wonder if we can control the depth shadow,the result may be OK.But I don't know how to remove the shadow in depth image.

[idata]

Hi Cindyisme,

I would like to investigate a little bit more regarding your query and as soon as I have useful information I'll let you know. Thanks for your patience during the meantime. Regards,-Yermi A.

[idata]

Hi Cindyisme,

Thanks for your patient. To lessen the shadow effect, try lowering the setting depth confidence threshold. In Windows SDK https://software.intel.com/sites/landingpage/realsense/camera-sdk/v2016r3/documentation/html/index.html?depthconfidencethreshold_device_pxccapture.html https://software.intel.com/sites/landingpage/realsense/camera-sdk/v2016r3/documentation/html/index.html?depthconfidencethreshold_device_pxccapture.html. In librealsense you can play with it using the cpp-config-ui sample. Hope this helps. Regards,-Yermi A.