Retrieving depth information from ZR300

[nnain1]

I used ZR300 for depth measurement.

I can't retrieve correct Depth matrix from the camera.

What I did was as follow and retrieve the depth matrix from the stream.

const uint16_t * depth_image = (const uint16_t *)dev->get_frame_data(rs::stream::depth);

float scale = dev->get_depth_scale();

float data[HEIGHT*WIDTH];

for(int dy=0; dy

{

for(int dx=0; dx

{

uint16_t depth_value = depth_image[dy * depth_intrin.width + dx];

if(depth_value == 0)

continue;

else

data[dy*dx] = (float)depth_value * scale;

}

}

Mat depthimage(HEIGHT, WIDTH, CV_32FC1, &data);

Mat colorimage(HEIGHT, WIDTH, CV_8UC3, (uchar*)dev->get_frame_data(rs::stream::color));

imshow("dst", depthimage);

waitKey(1);

dst.release();

I have another depth follow sample from realsense, that has correct object.

The code is

glPixelTransferf(GL_RED_SCALE, 0xFFFF * dev->get_depth_scale() / 2.8f);

glDrawPixels(WIDTH, HEIGHT, GL_RED, GL_UNSIGNED_SHORT, dev->get_frame_data(rs::stream::depth));

glPixelTransferf(GL_RED_SCALE, 1.0f);

Red color image is the one following sample program and the image "dst" is the one using the first code.

Another one is the color image, we can see there is a cabinet.

Why I don't have correct depth image, what is wrong with my code?

[idata]

Hello inmn,

Sorry for the late response. While we don't usually debug customer's code, I am trying to reproduce your issue but have encountered some obstacles. Please stay tuned...Regards,JesusIntel Customer Support

[idata]

Hello inmn,

It seems that you have to do a conversion of the pixel formats from RealSense to OpenCV. See the file https://github.com/IntelRealSense/realsense_sdk_zr300/blob/master/sdk/src/core/image/image_conversion_util.cpp for an example of how to do this.Let us know if this helps.Regards,JesusIntel Customer Support

[nnain1]

Hello Jesus,

Many thanks for the reply.

I have the same issue for color image. I tried to grab into Opencv's Mat image.

What I did was

dev.enable_stream(rs::stream::color, 640, 480, rs::format::bgr8, 60); cv::Mat colorImg(480, 640, CV_8UC3, (unsigned char *)dev.get_frame_data(rs::stream::color)); imshow("colorImg",colorImg); cv::waitKey(1);

Then my color image became. Is that the same issue as I need to pixel format conversion?

[nnain1]

Can I have any sample of usage?

[nnain1]

Thanks now ok.

I take the image as rgb8

dev.enable_stream(rs::stream::color, 640, 480, rs::format::rgb8, 60);

Then convert to RGBtoBGR

cv::cvtColor(srcMat, grayMat, CV_RGB2BGR);

It works.

[nnain1]

Hello Jesus,

Many thanks for the reply.

I have the same issue for color image. I tried to grab into Opencv's Mat image.

What I did was

dev.enable_stream(rs::stream::color, 640, 480, rs::format::bgr8, 60); cv::Mat colorImg(480, 640, CV_8UC3, (unsigned char *)dev.get_frame_data(rs::stream::color)); imshow("colorImg",colorImg); cv::waitKey(1);

Then my color image became. Is that the same issue as I need to pixel format conversion?