本文记录将 16 位的 深度图伪彩色化的代码。
注意,DepthMapPseudoColorize 函数第一个参数可以是 图像的 data,也可以是 path。
在函数内部都可以。
在第二篇参考文献中,有 不同的色彩参数。
import numpy as np
import cv2
from PIL import Image
def DepthMapPseudoColorize(depth_map,output_path, max_depth = None, min_depth = None):
# Load 16 units depth_map(element ranges from 0~65535),
# Convert it to 8 units (element ranges frome 0~255) and pseudo colorize
if isinstance(depth_map, np.ndarray):
uint16_img = depth_map
# Indicate the argument load mode -1, otherwise loading default 8 units
if isinstance(depth_map, str):
uint16_img = cv2.imread(depth_map, -1)
if None == max_depth:
max_depth = uint16_img.max()
if None == min_depth:
min_depth = uint16_img.min()
uint16_img -= min_depth
uint16_img = uint16_img / (max_depth - min_depth)
uint16_img *= 255
# cv2.COLORMAP_JET, blue represents a higher depth value, and red represents a lower depth value
# The value of alpha in the cv.convertScaleAbs() function is related to the effective distance in the depth map. If like me, all the depth values
# in the default depth map are within the effective distance, and the 16-bit depth has been manually converted to 8-bit depth. , then alpha can be set to 1.
im_color=cv2.applyColorMap(cv2.convertScaleAbs(uint16_img,alpha=1),cv2.COLORMAP_JET)
#convert to mat png
im=Image.fromarray(im_color)
im.save(output_path)
调用实例:
input_dir = 'D:/Application/NonAcademic/Desktop/MC_Result/4_0214MC_loc/'
input_file_name = '1024_4_depth.png'
output_dir = 'D:/Application/NonAcademic/Desktop/MC_Result/4_0214MC_loc/0215depth_colorized/'
output_file_name = input_file_name
input_path = input_dir+input_file_name
output_path = output_dir+output_file_name
DepthMapPseudoColorize(input_path,output_path)
[1] python 将16位 png 深度图转化为伪彩色图
[2] 深度图转伪彩色图(python)