lbp的matlab实现

lbp的matlab实现

1、基本lbpfunction lbpI = lbp(I) I = imresize(I,[256 256]); [m,n,h] = size(I); if
h==3 I = rgb2gray(I); end lbpI = uint8(zeros([m n])); for i = 2:m-1 for j =
2:n-1 neighbor = [I(i-1,j-1) I(i-1,j) I(i-1,j+1) I(i,j+1) I(i+1,j+1) I(i+1,j)
I(i+1,j-1) I(i,j-1)] > I(i,j); pixel = 0; for k = 1:8 pixel = pixel +
neighbor(1,k) * bitshift(1,8-k); end lbpI(i,j) = uint8(pixel); end end2、等价模式
function lbpI = lbp_equivalent(I) I = imresize(I,[256 256]); [m,n,h] = size(I);
if h==3 I = rgb2gray(I); end lbpI = uint8(zeros([m n])); table = lbp59table();
for i = 2:m-1 for j = 2:n-1 neighbor = [I(i-1,j-1) I(i-1,j) I(i-1,j+1) I(i,j+1)
I(i+1,j+1) I(i+1,j) I(i+1,j-1) I(i,j-1)] > I(i,j); pixel = 0; for k = 1:8 pixel
= pixel + neighbor(1,k) * bitshift(1,8-k); end lbpI(i,j) =
uint8(table(pixel+1)); end end %跳跃点 function count = getHopcount(i) i =
uint8(i); bits = zeros([1 8]); for k=1:8 bits(k) = mod(i,2); i =
bitshift(i,-1); end bits = bits(end:-1:1); bits_circ = circshift(bits,[0 1]);
res = xor(bits_circ,bits); count = sum(res); % lbp表 function table =
lbp59table() table = zeros([1 256]); temp = 1; for i=0:255 if getHopcount(i)<=2
table(i+1) = temp; temp = temp + 1; end end
3、旋转不变性
function lbpI = lbp_rotation(I) I = imresize(I,[256 256]); [m,n,h] = size(I);
if h==3 I = rgb2gray(I); end lbpI = uint8(zeros([m n])); for i = 2:m-1 for j =
2:n-1 neighbor = [I(i-1,j-1) I(i-1,j) I(i-1,j+1) I(i,j+1) I(i+1,j+1) I(i+1,j)
I(i+1,j-1) I(i,j-1)] > I(i,j); pixel = 0; for k = 1:8 pixel = pixel +
neighbor(1,k) * bitshift(1,8-k); end lbpI(i,j) = uint8(rotationMin(pixel)); end
end % 计算最小值 function minval = rotationMin(i) i = uint8(i); vals = ones([1
8])*256; for k=1:8 vals(k) = i; last_bit = mod(i,2); i = bitshift(i,-1); i =
last_bit * 128 + i; end minval = min(vals);

• « 上一篇：使用ImToken钱包，如何创建、导入、使用EOS钱包及账户
• » 下一篇：前端项目开发经验总结（持续更新）