-------------------------
***완전히 똑같은 두 이미지 input 으로 넣었을 때 각 평가지표 값***

MSE(Mean Squared Error):  0.0  
RMSE(Root Mean Squared Error):  0.0  
PSNR(Peak Signal-to-Noise Ratio):  inf  
SSIM(Structural Similarity Index):  (1.0, 1.0)  
UQI(Universal Quality Image Index):  1.0  
MSSSIM(Multi-scale Structural Similarity Index):  (1+0j)  
ERGAS(Erreur Relative Globale Adimensionnelle de Synthèse):  0.0  
SCC(Spatial Correlation Coefficient):  0.9972222222222222  
RASE(Relative Average Spectral Error):  0.0  
SAM(Spectral Angle Mapper):  0.0  
VIF(Visual Information Fidelity):  0.9999999999831712  

-------------------------

-------------------------
***사람이 보기에 비슷한 느낌의 '단순한' 이미지 input 으로 넣었을 때 각 평가지표 값***

흰배경 - 사과 vs 일본국기  
MSE(Mean Squared Error):  3086.728214449017  
RMSE(Root Mean Squared Error):  55.5583316384592  
PSNR(Peak Signal-to-Noise Ratio):  13.23581969209389  
SSIM(Structural Similarity Index):  (0.7796337234351008, 0.8114930770883629)  
UQI(Universal Quality Image Index):  0.8921791981212756  
MSSSIM(Multi-scale Structural Similarity Index):  (0.6931649863289358+0j)  
ERGAS(Erreur Relative Globale Adimensionnelle de Synthèse):  7079.446899828788  
SCC(Spatial Correlation Coefficient):  -0.003083131443304951  
RASE(Relative Average Spectral Error):  1035.5288707858879  
SAM(Spectral Angle Mapper):  0.22525911289791423  
VIF(Visual Information Fidelity):  0.014134211610046144  

-------------------------

-------------------------
***사람이 보기에 비슷한 느낌의 '복잡한' 이미지 input 으로 넣었을 때 각 평가지표 값***

스핑크스 vs 밀밭  
MSE(Mean Squared Error):  4228.241950160036  
RMSE(Root Mean Squared Error):  65.02493329608295  
PSNR(Peak Signal-to-Noise Ratio):  11.869205301480976  
SSIM(Structural Similarity Index):  (0.17587595405226664, 0.19594982461565502)  
UQI(Universal Quality Image Index):  0.8052137397560033  
MSSSIM(Multi-scale Structural Similarity Index):  (0.19348811664564788+0j)  
ERGAS(Erreur Relative Globale Adimensionnelle de Synthèse):  20921.526683233562  
SCC(Spatial Correlation Coefficient):  -0.007230694327466757  
RASE(Relative Average Spectral Error):  3034.279373635596  
SAM(Spectral Angle Mapper):  0.41485073929497895  
VIF(Visual Information Fidelity):  0.013447307963235655  

-------------------------

-------------------------
***사람이 보기에 전혀 다른 느낌의 이미지를 input 으로 넣었을 때 각 평가지표 값***

아그리파 vs 시내버스  
MSE(Mean Squared Error):  9474.102110196616  
RMSE(Root Mean Squared Error):  97.33499941026669  
PSNR(Peak Signal-to-Noise Ratio):  8.36542299701647  
SSIM(Structural Similarity Index):  (0.21393179837966025, 0.2725072131183044)  
UQI(Universal Quality Image Index):  0.6924552155374301  
MSSSIM(Multi-scale Structural Similarity Index):  (0.11500393747712011+0.05118862207347572j)  
ERGAS(Erreur Relative Globale Adimensionnelle de Synthèse):  40920.827193207966  
SCC(Spatial Correlation Coefficient):  -0.00024019532217466047  
RASE(Relative Average Spectral Error):  5903.915434761586  
SAM(Spectral Angle Mapper):  0.5708098319543257  
VIF(Visual Information Fidelity):  0.019232536770568145  

-------------------------

ImageSimilarityTest3.ipynb

Untitled

‘\’ 방향으로 밝아야 함

Untitled

‘\’ 방향으로 어두워야 함

image processing - average hashing 활용 (4주차 핵심기능 구현 예정)

Image hashing with OpenCV and Python - PyImageSearch

Source image

Untitled