# 红外非均匀性校正算法总结

Blog: https://blog.csdn.net/AnimateX

Email: animatex.deng@gmail

1. 基于定标的非均匀性校正

2. 基于场景的非均匀性校正

## 3 基于定标的非均匀性校正

$V\left(i,j,\varphi \right)={R}_{i,j}{\varphi }_{s}+{T}_{i,j}\phantom{\rule{2.8cm}{0ex}}\left(1\right)$

${V}_{L}\left(i,j,\varphi \right)={R}_{i,j}×{\varphi }_{L}+{T}_{i,j}\phantom{\rule{2cm}{0ex}}\left(2\right)$

${V}_{H}\left(i,j,\varphi \right)={R}_{i,j}×{\varphi }_{H}+{T}_{i,j}\phantom{\rule{2cm}{0ex}}\left(3\right)$

${Y}_{L}\left(i,j\right)={A}_{i,j}×{V}_{L}+{B}_{i,j}\phantom{\rule{2.8cm}{0ex}}\left(4\right)$

${Y}_{H}\left(i,j\right)={A}_{i,j}×{V}_{H}+{B}_{i,j}\phantom{\rule{2.8cm}{0ex}}\left(5\right)$

Code

function imgOut = Calibration( object, baff)
%Calibration just process one frame.
[M, N] = size( object );

imgOut = coeff .* (object - baff);

end
% 如果是高阶或者多点校正，参考下面说明(化简)
% Three point, Piecewise linear correction
%   V = A*(x^2) + B*x + C
%       V1 = A * (a1^2) + B * (a1) + C
%       V2 = A * (a2^2) + B * (a2) + C
%       V3 = A * (a3^2) + B * (a3) + C
%   Y = mean( V )= A * V^2 + B * V + C
%       Y1 = A * V1^2 + B * V1 + C
%       Y2 = A * V2^2 + B * V2 + C
%       Y3 = A * V3^2 + B * V3 + C
%   Use Cramer's Rule solve linear equations
%   Matrix: D D1 D2 D3
%       |D|  = (V2 - V1) * (V3 - V1) * (V3 - V2)
%       |D1| = mean(V1) * V2 + mean(V2) * V3 + mean(V3) * V1
%              - mean(V1) * V3 - mean(V2) * V1 - mean(V3) * V2
%       |D2| = [mean(V1)]^2 * mean(V2) + [mean(V2)]^2 * mean(V3)
%              + [mean(V3)]^2 * mean(V1) - [mean(V1)]^2 * mean(V3)
%              - [mean(V2)]^2 * mean(V1) - [mean(V3)]^2 * mean(V2)
%       |D3| = [mean(V1)]^2 * mean(V3) * V2 + [mean(V2)]^2 * mean(V1) * V3
%              + [mean(V3)]^2 * mean(V2) * V1 - [mean(V1)]^2 * mean(V2) * V3
%              - [mean(V2)]^2 * mean(V3) * V1 - [mean(V3)]^2 * mean(V1) * V2
%
%       A = |D1| / |D|
%       B = |D2| / |D|
%       C = |D3| / |D|
%

• 广告
• 抄袭
• 版权
• 政治
• 色情
• 无意义
• 其他

120