CH5

Neighborhood Processing

5-1 Introduction

linear filter:將灰階值進行線性函數運算

Spatial filtering thus reqires three steps:
(1)Position the mask over the current pixel
安置在當前指定像素的遮罩
(2)Form all products of filter elements with the corresponding elements of the neighborhood
fliter elements與corresponding elements相乘
(3)Add all the products
相加


5-2 Notation表記法

mask
超過圖片範圍時會產生陰影

Ignore the edges
忽略邊緣

Pad with zeros
被忽略邊緣補零


5-3 Filtering in MATLAB

filter2(filter,image,'same')補零的動作

filter2(filter,image,'valid')忽略邊緣

filter2(filter,image,'full')影像會比原圖還要大


5-4 Frequencies: Low-and High-Pass-Filters

(High-frequency components)：內部灰階值變化比較大
(Low -frequency components)：灰階值變化不大
(High-pass Filter)
高通濾波器----讓高頻率的部分通過
(Low-pass Filter)
低通濾波器----保留低頻率,平均濾波器為低通濾波器


5-5 Gaussian Filters 高斯濾波器

1.高斯函數透過傅立葉轉換後還是高斯函數
2.旋轉後都還是對稱的
3.分離性
4.兩個高斯函數的旋積是另一高斯函數