如何在Python中获取图像分割区域的平均RGB值并保存为CSV文件

Hey there! Let's walk through exactly how to get the average RGB values of a segmented region in your image, convert the relevant data to a numpy array, and save everything to a CSV. I'll break this down into simple, actionable steps:

1. Install Required Libraries #

First, make sure you have the tools we need. Open up your terminal and run:

pip install opencv-python numpy pandas

Or if you prefer using PIL/Pillow instead of OpenCV, install this set:

pip install pillow numpy pandas

2. Load Your Image & Convert to a Numpy Array #

We need to turn your image into a numpy array—this makes it easy to manipulate pixel data. Here are two common methods:

Option A: Using OpenCV #

import cv2
import numpy as np
import pandas as pd

# Load the image (note: OpenCV defaults to BGR format, so we'll convert to RGB)
img = cv2.imread("your_image_path.png")
img_rgb = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)

Option B: Using PIL/Pillow #

from PIL import Image
import numpy as np
import pandas as pd

# Load the image and convert directly to a numpy array (already in RGB)
img = Image.open("your_image_path.png")
img_rgb = np.array(img)

Pro tip: The resulting img_rgb array has a shape like (height, width, 3)—each element corresponds to a pixel's R, G, B values respectively.

3. Isolate Your Segmented Region #

This is the key part—you need to define which part of the image you're targeting. Here are two common scenarios:

Scenario 1: Rectangular Region (Manual Coordinates) #

If your segmented area is a rectangle, specify the y-range (top to bottom) and x-range (left to right):

# Example: Region from y=100 to y=300, x=200 to x=400
segmented_region = img_rgb[100:300, 200:400, :]

Scenario 2: Irregular Region (Using a Mask) #

If your segmentation is an irregular shape, use a mask image (where the target region is white, 255, and the rest is black, 0):

# Load the mask as a grayscale image
mask = cv2.imread("your_mask_path.png", 0)
# Extract only the pixels where the mask is white
segmented_region = img_rgb[mask == 255]

4. Calculate the Average RGB Values #

Now that we have our region, computing the average is straightforward:

# Calculate the mean for each RGB channel (axis=0 averages across all pixels)
avg_rgb = np.mean(segmented_region, axis=0)
# Convert to integers since RGB values are 0-255 whole numbers
avg_rgb = avg_rgb.astype(int)

print(f"Average RGB values for the region: R={avg_rgb[0]}, G={avg_rgb[1]}, B={avg_rgb[2]}")

5. Save Data to CSV #

You can save either all the pixel data from the region, or just the average values. Here's how to do both:

Save All Pixel RGB Data #

# Reshape the region array into a 2D format (each row = one pixel's RGB)
pixel_data = segmented_region.reshape(-1, 3)

# Option 1: Using Pandas (cleaner column names)
df = pd.DataFrame(pixel_data, columns=["R", "G", "B"])
df.to_csv("pixel_rgb_data.csv", index=False)

# Option 2: Using Numpy (no Pandas required)
np.savetxt(
    "pixel_rgb_data_numpy.csv",
    pixel_data,
    delimiter=",",
    header="R,G,B",
    comments="",
    fmt="%d"  # Save as integers
)

Save Only the Average RGB Value #

# Option 1: Using Pandas
avg_df = pd.DataFrame([avg_rgb], columns=["Avg_R", "Avg_G", "Avg_B"])
avg_df.to_csv("average_rgb.csv", index=False)

# Option 2: Using Numpy
np.savetxt(
    "average_rgb_numpy.csv",
    [avg_rgb],
    delimiter=",",
    header="Avg_R,Avg_G,Avg_B",
    comments="",
    fmt="%d"
)

Quick Notes to Avoid Headaches #

• Always double-check that your image is in RGB format (not BGR) before calculating averages—otherwise your colors will be off.
• For irregular regions, make sure your mask is properly aligned with your original image (same dimensions, correct region marked).
• Using fmt="%d" in numpy's savetxt ensures your RGB values are saved as whole numbers instead of decimals.

内容的提问来源于stack exchange，提问作者Thijs Brokking