显著性检测文献阅读笔记之FT算法

论文名称

Frequency-tuned Salient Region Detection

作者

Radhakrishna Achanta, Sheila Hemami, Francisco Estrada, and Sabine Süsstrunk

引用标注

R. Achanta, S. Hemami, F. Estrada and S. Susstrunk, Frequency-tuned Salient Region Detection, IEEE International Conference on Computer Vision and Pattern Recognition (CVPR), 2009.

算法思路

1. 读取图片
2. 对图片进行高斯滤波
3. 将图片从RGB色彩空间转换到Lab色彩空间，得到L(x, y)，A(x, y)，B(x, y)
4. 在Lab色彩空间上计算整幅图像的平均值Lm，Am，Bm
5. 计算显著图，原图每个像素点出的显著性计算公式为：S(x,y)=(L(x,y) - Lm)^2 + (A(x,y) - Am)^2 + (B(x,y) - Bm)^2

源码解读（文献作者的原版代码）

% 读取图片img = imread('test.jpg');% 对图片进行高斯滤波imgGausFilter = imfilter(img, fspecial('gaussian', 3, pi/2.75), 'symmetric', 'conv');% 将图片转换到Lab色彩空间（采用d65照明标准）cform = makecform('srgb2lab', 'AdaptedWhitePoint', whitepoint('d65'));imgLab = applycform(imgGausFilter,cform);% 取得图像的L,a,b分量L = double(imgLab(:,:,1));A = double(imgLab(:,:,2));B = double(imgLab(:,:,3));% 分别计算图像的L,a,b分量的平均值Lm = mean(mean(L));Am = mean(mean(A));Bm = mean(mean(B));% 分别计算每个像素点处的显著性saliencyMap = (L-Lm).^2 + (A-Am).^2 + (B-Bm).^2;% 显示显著图figure, imshow(saliencyMap,[]);