How to get xyz coordinates of pointcloud using Intel Real Sense D415?

edit

As soon as I had got your answer I started working to extract xyz coordinates and and I finally get it through this code(https://www.programmersought.com...

Broken link, can you please! fix this

but still struggling to get the max and min of xyz. If you don't mind can you plz elaborate how get this whether i used array or list function and then try to get the min and max or anything you might suggest.

It is an easy to trick you just have to check the distance from the camera to the detected object and calculate z means a distance of that object so you can get which object is near and which object is far, but before that obtain x,y,z from UV first.

Hello Ranjit, once again thank you so much for helping me. According to the link you have provided I have got code up to this, kindly help me out what mistakes I am doing 1. (First you have to subscribe to camera_info, pixel_coordinate, and point cloud topic.) I have subscribe the point cloud and camera_info topic but didn't get the pixel coordinate

def subscribe(self):
            self.pc_sub = rospy.Subscriber("/camera/depth_registered/points", PointCloud2, self.pc_callback)
            self.info_sub = rospy.Subscriber("/camera/rgb/camera_info", CameraInfo, self.info_callback)
            self.camera_image = rospy.Subscriber("") #How to subscribe pixel coordinate(what is the topic name)

   def info_callback(self, data):
            self.camera_model = image_geometry.PinholeCameraModel()
            self.camera_model.fromCameraInfo(data)
            self.info_sub.unregister()

2.(Call get_depth function for obtaining the distance of that pixel into the point cloud.)

def get_depth(self, x, y):
    gen = pc2.read_points(self.pc, field_names='z', skip_nans=False, uvs=[(x, y)])
    return next(gen)

3.Call camera_model.projectPixelTo3dRay((U,V)) 4.normalize Z and then after from that vector multiply get_depth function result(distance) with the vector X and Y value.

I didn't completely understand 3 and 4 point.Is I am doing right?

def calculate_3d_point(self, pixel):
        lookup = self.depth.load()
        depth = lookup[pixel[0], pixel[1]]  # lookup pixel in depth image
        ray = self.model.projectPixelTo3dRay(tuple(pixel))
        ray_z = [el / ray[2] for el in ray] 
        pt = [el * depth for el in ray_z] 
        point = PointStamped()
        point.header.frame_id = self.frame
        point.point.x = pt[0]
        point.point.y = pt[1]
        point.point.z = pt[2]
        return point

I didn't completely understand 3 and 4 point.Is I am doing right?

I think you are going right over here, but importing thing to note first you have to achieve point(U, V) value from your video stream of the particular detected object. then you can convert it to a 3D coordinate. In simple words, You need to have a 2D coordinate (cartesian coordinates) to obtain a 3D coordinate and processed it to point cloud form with PinholeCameraModel.

If you think that this answer is right, then press on tick button so in the future this answer can help others.

Thanks!

Thanks you so much Ranjit, I have answered the ques. In my application I have to calculate the volume of soil from pointcloud. kindly check it if anything still wrong? Thanks again.