2D Fast Marching Computations

Fast Marching method 跟 dijkstra 方法类似,只不过dijkstra方法的路径只能沿网格,而Fast Marching method的方法可以沿斜线.
[Level Set Methods and Fast Marching Methods p94-95

]

这里$u$理解为到达点的时间, $F_{ijk}$理解为在点$ijk$的流速.
然后就可以跟Boundary Value Formulation对应起来了.

[Level Set Methods and Fast Marching Methods p7

]

本例,首先加载一张灰度图片, 将里面像素的值W作为该点的流速F. 然后算到达该点的时间,也就是程序里面的距离D.

matlab部分源代码如下:
example1.m

n = 400;
[M,W] = load_potential_map('road2', n);
%start_point = [16;219];
%end_point = [394;192];
% You can use instead the function
[start_point,end_point] = pick_start_end_point(W);

clear options;
options.nb_iter_max = Inf;
options.end_points = end_point; % stop propagation when end point reached
[D,S] = perform_fast_marching(W, start_point, options);
% nicde color for display
A = convert_distance_color(D);
imageplot({W A}, {'Potential map' 'Distance to starting point'}); 
colormap gray(256);

perform_front_propagation_2d_slow.m

function [D,S,father] = perform_front_propagation_2d_slow(W,start_points,end_points,nb_iter_max,H)

%   [D,S] = perform_front_propagation_2d_slow(W,start_points,end_points,nb_iter_max,H);
%
%   [The mex function is perform_front_propagation_2d]
%
%   'D' is a 2D array containing the value of the distance function to seed.
%   'S' is a 2D array containing the state of each point : 
%       -1 : dead, distance have been computed.
%        0 : open, distance is being computed but not set.
%        1 : far, distance not already computed.
%   'W' is the weight matrix (inverse of the speed).
%   'start_points' is a 2 x num_start_points matrix where k is the number of starting points.
%   'H' is an heuristic (distance that remains to goal). This is a 2D matrix.
%   
%   Copyright (c) 2004 Gabriel Peyr?

data.D = W.*0 + Inf; % action 先把所有点的距离标为Inf
start_ind = sub2ind(size(W), start_points(1,:), start_points(2,:));
data.D( start_ind ) = 0; %将起点的距离设置为0
data.O = start_points; % 将起点加入Open list 
% S=1 : far,  S=0 : open,  S=-1 : close
data.S = ones(size(W));% 将所点的状态设为Far
data.S( start_ind ) = 0; % 将起点的状态设为open(trial)
data.W = W;
data.father = zeros( [size(W),2] );% father维度400*400*2,父节点有两个,因为走斜线

verbose = 1;

if nargin<3
    end_points = [];
end
if nargin<4
    nb_iter_max = round( 1.2*size(W,1)^2 );
end
if nargin<5
    data.H = W.*0;
else
    if isempty(H)
        data.H = W.*0;
    else
        data.H = H;
    end
end

if ~isempty(end_points)
    end_ind = sub2ind(size(W), end_points(1,:), end_points(2,:));
else
    end_ind = [];
end

str = 'Performing Fast Marching algorithm.';
if verbose
    h = waitbar(0,str);
end

i = 0; 
while i<nb_iter_max && ~isempty(data.O) && isempty( find( data.S(end_ind)==-1 ) )
    i = i+1;
    data = perform_fast_marching_step(data);
    if verbose
        waitbar(i/nb_iter_max, h, sprintf('Performing Fast Marching algorithm, iteration %d.', i) );
    end
end

if verbose
    close(h);
end

D = data.D;
S = data.S;
father = data.father;


function data1 = perform_fast_marching_step(data)%有多个起点也是一样的,只需将他们的距离都设为0即可

% perform_fast_marching_step - perform one step in the Fast Marching algorithm
%
%   data1 = perform_fast_marching_step(data);
%
%   Data is a structure that records the state before/after a step 
%   of the FM algorithm.
%
%   Copyright (c) 2004 Gabriel Peyr?

% some constant
kClose = -1;
kOpen = 0;
kFar = 1;

D = data.D; % action, a 2D matrix
O = data.O; % open list
S = data.S; % state, either 'O' or 'C', a 2D matrix
H = data.H; % Heuristic
W = data.W; % weight matrix, a 2D array (speed function)
father = data.father;

[n,p] = size(D);  % size of the grid,   n,p都为400

% step size
h = 1/n;

if isempty(O)%看开集是否为空
    data1 = data;
    return;
end

ind_O = sub2ind(size(D), O(1,:), O(2,:));%获取开集里面的顶点的索引

[m,I] = min( D(ind_O)+H(ind_O) ); %m里面是最小值,I里面是该最小值在被检测矩阵里面的索引
I = I(1);%取第一个索引
% selected vertex
% 取开集中的第I个点的索引
i = O(1,I);
j = O(2,I);
O(:,I) = [];  % pop from open ,将此点从开集中移除
S(i,j) = kClose; % now its close, 将此点加入闭集(known set)中

% its neighbor
% 准备遍历他的右,上,左,下的邻近点
nei = [1,0; 0,1; -1,0; 0,-1 ];

for k = 1:4

    ii = i+nei(k,1);
    jj = j+nei(k,2);

    if ii>0 && jj>0 && ii<=n && jj<=p

        f = [0 0];  % current father

        %%% update the action using Upwind resolution
        P = h/W(ii,jj);
        % neighbors values
        a1 = Inf;
        if ii<n
            a1 = D( ii+1,jj );
            f(1) = sub2ind(size(W), ii+1,jj);
        end
        if ii>1 && D( ii-1,jj )<a1
            a1 = D( ii-1,jj );
            f(1) = sub2ind(size(W), ii-1,jj);
        end
        a2 = Inf;
        if jj<n
            a2 = D( ii,jj+1 );
            f(2) = sub2ind(size(W), ii,jj+1);
        end
        if jj>1 && D( ii,jj-1 )<a2
            a2 = D( ii,jj-1 );
            f(2) = sub2ind(size(W), ii,jj-1);
        end
        if a1>a2    % swap to reorder
            tmp = a1; a1 = a2; a2 = tmp;
            f = f([2 1]);
        end
        % now the equation is   (a-a1)^2+(a-a2)^2 = P^2, with a >= a2 >= a1.
        % 书上95页公式为:(ux^2 + uy^2)^(1/2)=1/Fijk
        % u理解为到达点的时间,Fijk理解为在点ijk处的流速

        % 那么 ux = (a-a1)/(1/400), uy = (a-a2)/(1/400)
        % 所以方程变为:((a-a1)^2/(1/400))^2+((a-a2)^2/(1/400))^2 = (1/Wij)^2
        % 把1/400移到右边,则得P

        if P^2 > (a2-a1)^2%delta 大于0
            delta = 2*P^2-(a2-a1)^2;
            A1 = (a1+a2+sqrt(delta))/2;
        else%否则用dijkstra方法,沿着格子走,公式为:max|ux,uy|=1/Fijk
            % (a-a1) / (1/400) = 1 / Wij
            A1 = a1 + P;
            f(2) = 0;%将第2个父节点设为0
        end

        switch S(ii,jj)
            case kClose%闭集不用更新
                % check if action has change. Should not appen for FM
                if A1<D(ii,jj)
                    % warning('FastMarching:NonMonotone', 'The update is not monotone');
                    % pop from Close and add to Open
                    if 0        % don't reopen close points
                        O(:,end+1) = [ii;jj];
                        S(ii,jj) = kOpen;
                        D(ii,jj) = A1;
                    end
                end
            case kOpen%开集才更新
                % check if action has change.
                if A1<D(ii,jj)
                    D(ii,jj) = A1;
                    father(ii,jj,:) = f;
                end
            case kFar%远集不仅更新,而且加入开集
                if D(ii,jj)~=Inf
                    warning('FastMarching:BadInit', 'Action must be initialized to Inf');  
                end    
                % add to open
                O(:,end+1) = [ii;jj];
                S(ii,jj) = kOpen;
                % action must have change.
                D(ii,jj) = A1;
                father(ii,jj,:) = f;
            otherwise
                error('Unknown state');
        end
    end
end

data1.D = D;
data1.O = O;
data1.S = S;
data1.W = W;
data1.H = H;
data1.father = father;

运行结果:

matlab完整源代码