Segmentation of the Pre-Processed Image Using the Nonparametric Confidence Interval Algorithm Using the Kernel Density Estimator

Abstract

The kernel density estimator is a nonparametric method to estimate the probability density function for any variable. Therefore, it is a statistical technique used to smooth each point in the data of the variable to be studied. As for nonparametric confidence intervals, they are also one of the estimation methods, which determines an interval that contains a set of values based on the sample data and is defined by an upper and lower bound. In this research, nonparametric confidence intervals were used to treat the noisy image with Gaussian noise, an pre-processing where the noise was removed from the image and filtered from the noise to give a clearly defined image free of noise, then the pre-processed image was segmented using the confidence interval method of the kernel density estimator, which smoothed the image data, as the important features of the image were identified by separating homogeneous areas. It was noted that the processed images on which the segmentation process was performed were better in showing Features from segmented images that contain noise. It was concluded that the bandwidth parameter, which controls the amount of data homogeneity, has an impact on the work of both the nonparametric confidence interval method and the kernel estimator. Also, in the pre-processing process, the Epanechnikov function was chosen because it gives better and clearer features than others.