Analisis Estimasi Kesalahan Citra Terdegradasi Noise Dengan Metode Statika MEAN SQUARE ERORR (MSE)

Abstract

In image processing sometimes to test an algorithm to reduce noise (noise), the noise is generated from the process of generating noise by using random numbers (uniform distribution). This study proposes the analysis of noise degraded images by measuring errors with the statistical method of the Mean Squared Error (MSE) estimation approach. The estimated results of MSE values ​ ​ on Gaussian noise degraded images show that the greater the mean and variance, the blurred the image. With a large variance, the image will experience as if only visible white image. The estimated MSE value with a variance of 0.05 obtained an average MSE value while the variance value of 0.50 MSE value equals zero. Furthermore, the estimated value obtained in the image of salt & pepper noise degraded is the greater the density value given, the MSE value also increases. Image display is also getting worse or blurred. Whereas the estimated value of MSE in Speckle noise degraded images is that the greater the given variance value will also result in the greater MSE value as well. The greater the value of the variance, the image will be more blurred.