Alright, let's tackle this question head-on. First, I'll walk you through the mathematical basics of calculating mean and standard deviation for a grayscale pixel frequency distribution, then show you exactly how to implement this in MATLAB—both with built-in functions (for speed) and manual calculations (for clarity on the math).

Grayscale images use values from 0 (black) to 255 (white). The frequency distribution here refers to how often each grayscale value appears across all pixels.

Mean Calculation #

The mean (average grayscale value) is a weighted average where each grayscale value is multiplied by its relative frequency (i.e., how many pixels have that value divided by total pixels). The formula looks like this:
$$\mu = \sum_{i=0}^{255} i \times p(i)$$
Where:

• $i$ = grayscale value (0 to 255)
• $p(i)$ = relative frequency of $i$ (count of $i$ / total pixels)

Standard Deviation Calculation #

Standard deviation measures how spread out the grayscale values are around the mean. First calculate the variance (average of squared differences from the mean), then take the square root. You can compute it two ways:

1. Directly using squared differences:
   $$\sigma = \sqrt{\sum_{i=0}^{255} (i - \mu)^2 \times p(i)}$$
2. A more computationally efficient version (avoids subtracting the mean from every value first):
   $$\sigma = \sqrt{\left(\sum_{i=0}^{255} i^2 \times p(i)\right) - \mu^2}$$

MATLAB has built-in tools to make this easy, but I'll cover both quick built-in methods and manual calculations so you understand the underlying process.

Option 1: Use Built-in Functions (Fast & Simple) #

If you just need the results without digging into the frequency distribution details, MATLAB's mean2 and std2 functions do all the work for you. These operate directly on the image matrix, which is mathematically identical to calculating from the frequency distribution.

% Step 1: Load and preprocess the image
img = imread('your_image_file.png'); % Replace with your image path
if size(img, 3) == 3 % Convert color images to grayscale
    img = rgb2gray(img);
end

% Step 2: Calculate mean and standard deviation
gray_mean = mean2(img); % Mean grayscale value
gray_std = std2(img); % Sample standard deviation (divides by N-1)

% If you need population standard deviation (divides by N), adjust it:
total_pixels = numel(img);
gray_std_pop = gray_std * sqrt((total_pixels - 1)/total_pixels);

Option 2: Manual Calculation (For Understanding the Math) #

If you want to work directly with the frequency distribution (histogram) of grayscale values, here's how to compute mean and std manually:

% Step 1: Load and preprocess the image (same as above)
img = imread('your_image_file.png');
if size(img, 3) == 3
    img = rgb2gray(img);
end

% Step 2: Get the grayscale histogram (counts per value)
[counts, bins] = imhist(img); % counts = pixel count per bin; bins = grayscale values (0-255)
total_pixels = numel(img);

% Step 3: Calculate relative frequencies
freq = counts / total_pixels;

% Step 4: Compute mean manually
mean_manual = sum(bins .* freq);

% Step 5: Compute standard deviation manually (using efficient formula)
variance_manual = sum(bins.^2 .* freq) - mean_manual.^2;
std_manual = sqrt(variance_manual); % This is population standard deviation

Key Notes #

• Always convert color images to grayscale first with rgb2gray—otherwise, you'll be calculating stats across RGB channels instead of grayscale values.
• If your image is stored as uint8 (the default for most images), MATLAB handles the calculations safely, but converting to double with img_double = double(img) can prevent overflow if you're doing more complex operations.
• The manual and built-in methods will give identical results (for population std)—the built-in functions are just optimized for speed, especially with large images.

Hope this clears things up—feel free to ask if you run into any issues with the code or need further clarification!

内容的提问来源于stack exchange，提问作者sara