【数字图像处理与应用】Lecture 4 图像增强(频域滤波)

数字图像处理与应用 Class 4. 20200506

Lecture 4 Image Enhancement: Filtering in the Frequency Domain

The Fourier series & the Fourier transform

DFT Properties

Steps of Filtering in the Frequency Domain

Some Basic Frequency Domain Filters

    Image smoothing

    Image sharpening

---

Introduction to Fourier Transform <!--傅里叶变换简介-->

The Big Idea

任何周期性重复自身的函数都可以表示为不同频率的正弦和余弦的和乘以不同的系数——傅里叶级数
即使是非周期函数，也可以表示为sin和/或cos的积分乘以一个权重函数。这里的公式是傅里叶变换。

通过傅里叶变换对函数进行变换，可以通过逆过程进行完全重构(恢复)，且不损失信息。

Introduction to Fourier Transform

(u) is the frequency variable.由上式积分可知，F(u)是由无穷个sin项和cos项之和组成，u的每个值决定了对应的sin -cos项的频率。

The Discrete Fourier Transform (DFT)

<!--离散傅里叶变换-->

Note： the coefficient 1/N also can be put in the first equation, or in both equations, or just keep the product of the two coefficients is equal to 1/N.

How to calculate the DFT

Step1: Let u=0, summing for all value of x

Step2: Substitute u=u+1, and repeat the summation over all value of x.

Step 3: If u=N-1, terminate, otherwise, return to step 2.

• $ N^2 $ summations and multiplications

More<