The range of pixel value when using Y8 and Y16 format.

[fwf-lernen]

Required Info
Camera Model	D455
Firmware Version	05.12.11.00
Operating System & Version	Ubuntu 16.04
Kernel Version (Linux Only)	4.15.0
Platform	PC
SDK Version	2.42.0
Language	C

Issue Description

When I use infrared camera to capture images in Y8 format, no matter how I reduce exposure time(neither using realsense-viewer nor call API: sensor.set_option(RS2_OPTION_EXPOSURE, exposure_time) ), I can't get pixel value below 15 even though I set the exposure time to 1usec.
The same thing happens when using RGB camera to capture images in Y16 format (min pixel value is 16*256 ). Besides, no matter how I increase exposure time, I can't get pixel value above 235*256.
Is the pixel value range set to 16 ~ 255 in Y8 and 16*256 ~ 235*256 in Y16? If not, how can I get pixel value that exceed the limit?

[MartyG-RealSense]

Hi @fwf-lernen I have researched your question carefully. Before moving forward though, could you provide further information about your meaning when you talk about the 'pixel value' to confirm that we are both thinking about the same thing?

Also, can you explain what you are trying to achieve that requires the setting that you are aiming for please? Thanks very much.

[fwf-lernen]

Hi @MartyG-RealSense, thank you for your reply.
`
// include the librealsense C++ header file
#include <librealsense2/rs.hpp>

// include OpenCV header file
#include <opencv2/opencv.hpp>

using namespace std;
using namespace cv;

int main()
{
//Contruct a pipeline which abstracts the device
rs2::pipeline pipe;

//Create a configuration for configuring the pipeline with a non default profile
rs2::config cfg;

//Add desired streams to configuration
cfg.enable_stream(RS2_STREAM_INFRARED, 640, 480, RS2_FORMAT_Y8, 30);

//Instruct pipeline to start streaming with the requested configuration
pipe.start(cfg);

// Camera warmup - dropping several first frames to let auto-exposure stabilize
rs2::frameset frames;
for(int i = 0; i < 30; i++)
{
    //Wait for all configured streams to produce a frame
    frames = pipe.wait_for_frames();
}

//Get each frame
rs2::frame ir_frame = frames.first(RS2_STREAM_INFRARED);

// Creating OpenCV matrix from IR image
Mat ir(Size(640, 480), CV_8UC1, (void*)ir_frame.get_data(), Mat::AUTO_STEP);
imwrite("ir_image.png", ir);

return 0;

}

`
Use this code to capture a image in Y8 format with a infrared camera , then just use " cv::Mat img = cv::imread("ir_image.png", IMREAD_GRAYSCALE) " to read the image. All values of the Mat are above 15, but I want to get values below 15. However, I can't do this by decreasing the exposure time.

[MartyG-RealSense]

Thanks very much for the information. I thought that you may have been talking about a cv::mat when you mentioned a range of 0 to 255 but wanted to make sure. I appreciate the confirmation.

The link below describes how OpenCV has a mechanism that can change the pixel value to a custom-defined value if it is less or greater than a certain amount. So for example, if the 'less than' threshold is set at 17 then that code could set that pixel to a lower value such as 1 or 0 when the pixel value is 16 and triggers the mechanism. Would that be a practical solution for you?

https://stackoverflow.com/questions/37342716/set-every-pixel-of-a-mat-to-a-certain-value-if-it%C2%B4s-lower-than-a-value

[fwf-lernen]

Thanks very much for your suggestion. But what I actually want to do is to calibrate the camera's inverse response function. So if I just set the value below 17 to 1 or 0, that means I can't calibrate the pixel value between 0 and 16. Honestly, if I can't get pixel value below 16, that means I needn't calibrate these values. I am here to find out why I can't get values below 16 directly. Is the camera set to be like this?

[MartyG-RealSense]

I ran the testing again in the Viewer in infrared mode with exposure at '1' and observed that even with gain minimized at the lowest allowed value of 16, detail could be faintly seen in the darkened image as the camera is moved around. This suggests to me that as long as gain cannot be reduced below '16' then the '16' pixel-values in your results may represent the faintly visible detail on the IR image.

A RealSense user in the case linked to below was also seeking to reduce gain below 16. The possibility of applying a physical optical filter over the camera lenses on the outside of the camera was suggested to them.

#2875 (comment)

[fwf-lernen]

OK, I think I have got what I want. Thanks again!
By the way, should I close the issue or you will do it? If you can, please close it. (I am going to sleep and can't reply to you in time.)

[MartyG-RealSense]

You are very welcome. :) I can close the case for you.