55 / 2016-12-22 18:36:37

Parametric Study on Deep Belief Network based Image Enhancement Algorithm for Pelvic Lesions Ultrasound Images

ultrasound,enhancement,noise,deep belief network

It is well-known that for analysis and interpretation, ultrasound images are the sources of information with cost effective and disclose the usage of hassle free environment. When an image of information is converted from one form to another using a commonly available process such as digitizing, scanning, transmitting, storing, the performance of the transformed image gets degraded up to some extent. Thus it is found that the quality of transformed image can be improved with the process of image enhancement technique. Consequently, very few methods are developed in diagnosing the lesions on ultrasound modality. Inspired by a neural network (NN), a system for image quality enhancement using the deep belief neural network is proposed. The statistical characteristics of image textures are determined from the noisy environment using the impact of entropy and autocorrelation features. In addition to that, the impact of skewness factor on image assessment is measured using the alignment problem to be considered. Finally, for obtaining the superiority of the enhanced image, the proposed method is successfully compared with the existing NN method