第四章亮度变化与滤波解析.ppt

* * Ex: an image [3 5 8;2 8 4;1 9 3] is given Output image is [T(3), T(5), T(8); T(2), T(8), T(4); …] Ex: T(f(x,y))=f(x,y)-f(x,y-1) ? neighborhood of (x,y) is {(x,y-1)} Filter h=[1, -1] * Ex: an image [3 5 8;2 8 4;1 9 3] is given Output image is [T(3), T(5), T(8); T(2), T(8), T(4); …] Ex: T(f(x,y))=f(x,y)-f(x,y-1) ? neighborhood of (x,y) is {(x,y-1)} Filter h=[1, -1] * * A curve with slope 1 has the effect of spreading gray levels … compressing Use histogram to explain Figure 3.5 (0~1.5*10^6)?0~6.2 * A curve with slope 1 has the effect of spreading gray levels … compressing Use histogram to explain Figure 3.5 (0~1.5*10^6)?0~6.2 Convolution vs Correlation Implementation Using IPT in MATLAB Imfilter (Try it!) Linear Smoothing Spatial Filters It has the effect of blurring image Noise can be reduced Examples Mean filter/box filter/averaging filter Gaussian filter/weighted filter Filter size 3×3 Filter size 9×9 Filter size 5×5 Filter size 15×15 Filter size 35×35 Original 15×15 Mean Filtering 函数imfilter 函数imfilter用来实现线性空间滤波。 例3.7 使用函数imfilter 3.4.2 非线性滤波器 非线性滤波器的定义 使用模板进行结果像素值的计算，结果值直接取决于像素邻域的值，而不使用乘积和的计算 R = w1z1 + w2z2 + … + wnzn 主要非线性滤波器 主要非线性滤波器 中值滤波 主要用途：钝化图像、去除噪音 计算公式：R = mid {zk | k = 1,2,…,9} 最大值滤波 主要用途：寻找最亮点 计算公式：R = max {zk | k = 1,2,…,9} 最小值滤波 主要用途：寻找最暗点 计算公式：R = min {zk | k = 1,2,…,9} 最大值滤波 最小值滤波 函数colfilt 函数colfilt Non-linear Smoothing Spatial Filters Median filter The best known example Special order-statistics filter Replacing the value of a pixel by the median of the gray levels in the predefined neighborhood (including the pixel being processed) of that pixel Median Median of a set of odd number values (assume the number is 2n+1) is the value such that n of them are greater or equal to it and n of them are less or equal to it Examples For a 3×3 neighborhood, the medi