Image Quality Evaluation Index
Related literature: Image quality assessment: from error visibility to structural similarity
For applications where images are ultimately viewed by humans, the only "correct" way to quantify visual image quality is through subjective assessment. In practice, however, subjective assessments are often too inconvenient, time-consuming, and expensive. The goal of objective image quality assessment research is to develop quantitative metrics that can automatically predict perceived image quality.
1.MSE(Mean Squared Error）
Mean square error, calculate the mean square value of the pixel difference between the original image and the distorted image, and then determine the degree of distortion of the distorted image by the size of the mean square value. The formula is as follows:
where M and N are the length and width of the image, fij represents the pixel value of the original image, and f'ij represents the pixel value of the degraded image.
2.PSNR（Peak Signal-to-Noise Ratio）
The peak signal-to-noise ratio, that is, the ratio of the energy of the peak signal to the average energy of the noise, is usually expressed in decibels (dB), and the general value range is 20-40. The larger the value, the smaller the distortion.
PSNR is an engineering term that represents the ratio of the maximum possible power of a signal to the destructive noise power that affects its representation accuracy. It is used as a measure of signal distortion and is often used as a measurement method for signal reconstruction quality in fields such as image compression.
PSNR is the most common and widely used image objective evaluation index, but it is based on the error between corresponding pixels , that is, image quality evaluation based on error sensitivity. Since the visual characteristics of the human eye are not taken into account (the human eye is more sensitive to contrast differences with lower spatial frequencies, and the human eye is more sensitive to brightness contrast differences than chromaticity, the human eye's perception of an area will be affected by The impact of its surrounding adjacent areas, etc.), so the evaluation results are often inconsistent with people's subjective feelings.
3.SSIM（structural similarity）
Structural similarity, the formula is based on three comparison measures, including luminance (luminance), contrast (contrast) and structure (structure).
General value range: 0-1. The larger the value, the better the image quality.
The basic flow of the algorithm block diagram is as follows
: for the input x and y, first calculate the luminance measurement, compare them, and get the first similarity-related evaluation; after subtracting the influence of luminance, calculate the contrast measurement, compare them, and get the second Two evaluations; use the results of the previous step to remove the contrast, and compare the structure. Finally, combine the results to get the final evaluation result
. From the perspective of implementation, the brightness is represented by the mean value, the contrast is represented by the variance after normalization by the mean value, and the structure is represented by the correlation coefficient (that is, r in the statistical sense, the product of covariance and variance) ratio).
Reference
1. Image quality evaluation index SSIM/PSNR/MSE
2. Image quality evaluation index: MSE, PSNR, SSIM