after align rgb to depth，the points calculated from depth image are not evenly distributed but a small pile of small

tguan · ‎03-15-2019

tguan · ‎03-15-2019

i calculate the points from aligned depth image,and it is devided into pieces,but if i do not align the depth image to color,it seems to be good.

__author__ = '关彤'
#encoding:utf-8
#RGB-D图片和点云数据
import pyrealsense2 as rs
import numpy as np
import cv2
 
pc = rs.pointcloud()
points = rs.points()
# Create a pipelinett
pipeline = rs.pipeline()
config = rs.config()
config.enable_stream(rs.stream.depth, 640, 480, rs.format.z16, 30)
config.enable_stream(rs.stream.color, 640, 480, rs.format.bgr8, 30)
# Start streaming
profile = pipeline.start(config)
# Getting the depth sensor's depth scale
depth_sensor = profile.get_device().first_depth_sensor()
depth_scale = depth_sensor.get_depth_scale()
print("Depth Scale is: " , depth_scale)
# 移除背景,设置阈值
clipping_distance_in_meters =1 #1 meter
clipping_distance = clipping_distance_in_meters / depth_scale
align_to = rs.stream.color
align = rs.align(align_to)
# Streaming loop（串流循环）
 
while True:
    frames = pipeline.wait_for_frames() # Get frameset of color and depth
    depth = frames.get_depth_frame()
    color = frames.get_color_frame()
    if (not depth) or (not color):
        break
    # Align the depth frame to color frame
    aligned_frames = align.process(frames)
 
    # Get aligned frames
    depth_frame = aligned_frames.get_depth_frame() # aligned_depth_frame is a 640x480 depth image
    depth_intrin = depth_frame.profile.as_video_stream_profile().intrinsics
    color_frame = aligned_frames.get_color_frame()
    # Validate that both frames are valid
    if not depth_frame or not color_frame:
        continue
#点云
    # Tell pointcloud object to map to this color frame
    #pc.map_to(color_frame)
    # Generate the pointcloud and texture mapping
    points = pc.calculate(depth_frame)
    testpoints = pc.calculate(depth)
 
    depth_image = np.asanyarray(depth_frame.get_data())
    color_image = np.asanyarray(color_frame.get_data())
 
 
    depth_image2 = np.where((depth_image > clipping_distance) | (depth_image <= 0), 0, depth_image)
    depth_image2 = cv2.applyColorMap(cv2.convertScaleAbs(depth_image2, alpha=0.03), cv2.COLORMAP_BONE)#渲染图像depth_colormap
    depth_image = cv2.applyColorMap(cv2.convertScaleAbs(depth_image, alpha=0.03), cv2.COLORMAP_BONE)
  
    cv2.namedWindow('Align Example', cv2.WINDOW_AUTOSIZE)
 
    cv2.imshow('Align Example', color_image)
  
    cv2.imshow('Align Example2', depth_image2)
    key=cv2.waitKey(1)
    if key == ord('q'):
      
        points.export_to_ply("point3.ply", color_frame)
        testpoints.export_to_ply("point0.ply", depth)
        break
 
pipeline.stop()

tguan · ‎03-15-2019

the two kind of points are like this

Eliza_D_Intel · ‎03-15-2019

Hello tguan, Thank you for your interest in the Intel RealSense Technology. Could you please let us know how your output looks like? Thank you and best regards, Eliza

Eliza_D_Intel · ‎03-22-2019

Hello tguan, Could you please let us know if you require further assistance on this subject? Thank you, Eliza