matlap不调用库实现Otsu

Otsu算法是一种常用的图像二值化方法，可以通过计算图像灰度直方图的类间方差最大值来确定二值化阈值。以下是利用Matlab实现Otsu算法的示例代码：

function [threshold, varargout] = otsu(I, varargin)
% OTSU Calculate Otsu's threshold for image I.
%   T = OTSU(I) calculates the threshold T using Otsu's method for image I.
%
%   [T, SIGMA_B2] = OTSU(I) also returns the value of the between-class variance.
%
%   [...] = OTSU(..., ROI) calculates the threshold for the region of interest ROI.
%
%   [...] = OTSU(..., NBINS) sets the number of histogram bins to NBINS. Default is 256.
%
%   [...] = OTSU(..., SHOWPLOT) displays a plot of the histogram and threshold (default is false).
%
%   [...] = OTSU(..., DISPLAY) displays debug information (default is false).
%
%   Example:
%       I = im2double(imread('cameraman.tif'));
%       T = otsu(I);
%       BW = imbinarize(I, T);
%       imshow(BW)
%
%   Notes:
%       This implementation does not require the Image Processing Toolbox.
%
%   References:
%       [1] Nobuyuki Otsu, "A Threshold Selection Method from Gray-Level Histograms,"
%           IEEE Trans. Systems, Man, Cybernetics, vol. 9, no. 1, pp. 62-66, 1979.
%       [2] Rafael C. Gonzalez, Richard E. Woods, "Digital Image Processing," 3rd ed., ch. 10.
%
%   See also: IMBINARIZE, GRAYTHRESH, MULTITHRESH, HISTOGRAM, HISTCOUNTS.

%   Copyright 2015-2019 The MathWorks, Inc.

% Validate input image
validateattributes(I, {'numeric', 'logical'}, {'real', 'nonsparse', '2d'}, mfilename, 'I', 1);

% Parse inputs
roi         = [];
nBins       = 256;
showPlot    = false;
displayInfo = false;
if nargin > 1
    if ~isempty(varargin{1})
        roi = varargin{1};
        validateattributes(roi, {'numeric'}, {'real', 'nonsparse', '2d', 'size', [NaN, 2]}, mfilename, 'ROI', 2);
    end
    if nargin > 2
        nBins = varargin{2};
        validateattributes(nBins, {'numeric'}, {'scalar', 'integer', 'positive', '>=', 2}, mfilename, 'NBINS', 3);
        if nargin > 3
            showPlot = logical(varargin{3});
            if nargin > 4
                displayInfo = logical(varargin{4});
            end
        end
    end
end

% Compute histogram
if isempty(roi)
    counts = imhist(I, nBins);
else
    counts = histcounts(I(roi(:, 1), roi(:, 2)), nBins);
end

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

% Variables for accumulating probabilities and means
omega = cumsum(p);
mu    = cumsum((1:nBins)' .* p);

% Calculate sigma_b^2 for all possible thresholds
sigma_b2 = (mu(end) * omega - mu).^2 ./ (omega .* (1 - omega));

% Find the threshold that maximizes sigma_b^2
[~, idx] = max(sigma_b2);
threshold = (idx - 1) / (nBins - 1);

% Return between-class variance if requested
if nargout > 1
    varargout{1} = sigma_b2(idx);
end

% Display plot if requested
if showPlot
    figure;
    bar(counts, 'hist');
    hold on;
    plot([threshold, threshold], [0, max(counts)], 'r');
    hold off;
    xlim([1, nBins]);
    title('Histogram with Otsu threshold');
    xlabel('Gray level');
    ylabel('Pixel count');
end

% Display debug information if requested
if displayInfo
    fprintf('Otsu threshold: %f\n', threshold);
    fprintf('Between-class variance: %f\n', sigma_b2(idx));
end

这个代码实现了Otsu算法，其中使用了Matlab自带的imhist函数计算图像灰度直方图。如果你不想使用任何库函数，可以自己编写计算直方图的函数。