Different Objective Image Quality Assessment Techniques

Quality assessment plays an improle in image analysis.The study highlight wavelet transformation based technique which is basically a time frequency transformation and compares it with traditional methods like PSNR, MSE, SSIM, MSSIM, UQI, NO REF BASED. By applying the wavelet decomposition to the image.The low-freq. components and the high-frequency components of the image are successively obtained .The bases of the wavelet decomposition can be selected from Haar, Daubechies (N=3), and Daubechies (N=5). The original image is reconstructed from the low-freq. and high-freq. components. This procedure is called the inverse wavelet transform. By reconstructing the image only from high frequency component, high frequency image is obtained and this image can be used for edge detection.Daubechies wavelet metric ((DWM) uses the concept of sharpness and zero crossing .Zero crossing provide location of sharp signal variation. Sharpness is defined by the boundaries between zones of different tones or colors.Four bands are extracted from reference and distorted image by employing 2-D Daubechies wavelet decomposition namely: LL, LH, HL, and HH.Sharpness of both reference and distorted can be calculated from energy in wavelet sub bandand then sharpness similarity of both original and distorted image is measured.By calculatingnumber of edge points in original and distorted image edge structural similarity is measured.Zero crossing is found from edge structural similarity.By combining zero crossing and structural similarity DWM is obtained. We performed our experiment on 9 publicly available images on which blurring of different types JPEG compression and JPEG compression with blurring are applied. For these images Daubechies wavelet metric(DWM) and traditional metrics like MSE,PSNR,SSIM,MSSIM ,UQI,NO REF BASED are obtained and results are compared .By comparing these results it is found that DWMgives comparable result to MSSIM and is better correlated with subjective scores