FFT is used to estimate the image resampling factor after interpolation

stay JPEG Detect linear and cubic interpolation in compressed images

Detection of Linear and Cubic Interpolation in JPEG Compressed Images

notes ：matlab Need to configure jpegtbx_1.4
Source download ：https://download.csdn.net/download/yohnyang/86249304

** Known existing image size 128*128, And the scope is 1.1-1.9 Between 9 Different tensile coefficients , After stretching the existing image differently , The test turns the image to the frequency domain to resample the stretched image ( Tensile coefficient ) It is estimated that , The algorithm estimates the variance of the second-order differential signal for each line of the image , Then it is Fourier transformed and mapped to the frequency domain , in the light of DFT The position of the spike in the signal estimates the resampling factor .**

 Algorithm flow ：
For  All images 
For  All stretching factors 
     The image is stretched and cubic Interpolation method 
     Image quantization 
     Find the second derivative 、 Standard deviation 
     The Fourier transform , Look for peaks in the spectrum , Find out the corresponding frequency of the peak 
     Estimate the number of stretched words by frequency , And calculate the estimation error , Record the estimated correct quantity 
End
End
 Calculate the probability of correct estimation 

in the light of DFT It is easy to misjudge when searching for peak , This algorithm applies neighborhood peak searching , Calculate the reference frequency through the known stretching factor , Then set the threshold for T The frequency is in the neighborhood of the reference frequency [-T,T] Search the precise frequency calculation value , As shown in the figure below , Algorithm application T=5, many