大津阈值分割算法matlab

以下是大津阈值分割算法的MATLAB代码：

function [bw, threshold] = otsu(im)
%OTSU performs Otsu's thresholding on a grayscale image
%   [BW, THRESHOLD] = OTSU(IM) performs Otsu's thresholding on the
%   grayscale image IM and returns the binary image BW and the
%   threshold value THRESHOLD.

% Ensure input is a grayscale image
if size(im, 3) > 1
    error('Input must be a grayscale image');
end

% Compute histogram
counts = imhist(im);

% Normalize histogram
p = counts / sum(counts);

% Compute cumulative sums
omega = cumsum(p);
mu = cumsum(p .* (1:length(p))');

% Compute global mean
mu_t = mu(end);

% Compute between-class variance
sigma_b_squared = (mu_t * omega - mu).^2 ./ (omega .* (1 - omega));

% Find maximum between-class variance
[threshold, ~] = max(sigma_b_squared);

% Convert threshold to index
threshold = threshold / numel(p);

% Binarize image using threshold
bw = imbinarize(im, threshold);
end

该函数的输入为一个灰度图像，输出为二值图像和阈值值（即otsu算法计算出的最佳阈值）。函数首先计算图像的直方图，然后将其归一化。接下来，该函数计算每个像素值的概率分布。然后，它计算每个像素值的累计和，并使用这些值计算全局平均值和类间方差。最后，该函数找到类间方差最大的阈值，并使用该阈值将图像二值化。