DEVELOPMENT AND IMPLEMENTATION OF ALGORITHM FOR IMAGE PREPROCESSING OF MICROORGANISM

The digital revolution has changed most aspects of modern life. Nowhere has the change been more fundamental than in the field of microscopy. Researchers who use the microscope in their investigations have been among the pioneers who applied digital processing techniques to images. Vision is most powerful of the five senses of human being. Digitized visual information provides high impact on the subject. Digital image processing is concerned with the extraction of useful information from images. Visual information from microscopic images of microorganisms is analyzed regularly. This has resulted in a need to understand and implement digital processing on microscopic images. The purpose of this thesis is to bring new digital image processing techniques for the noise removal of microscopic image of microorganisms. The digitized image processing includes image representation; improving image quality by removing noise; and enhancing the quality of microscopic images. At the outset, the thesis elaborates on the concepts around microscopic images and their digital image processing. Various existing algorithms are studied for their efficacy. This thesis gives three different techniques of image processing based on the noise level in microscopic images. The thesis develops the techniques of image processing through “Simulation”, which is well accepted tool in the field of engineering. MATLAB has been used in this study to simulate the image processing algorithms. The algorithms developed in the study will be helpful in everyday life through better analysis of microscopic images of microorganisms. The thesis is a contribution to the medical field with better analytical techniques. This research work overviews different image processing techniques used in the analysis of microscopic images and other type of images. After reviewing, use of microscopic imaging is presented. Special emphasis is on two types of noise called Gaussian noise and Impulsive noise is given. In particular noise removal techniques are presented which will improve the visibility of microscopic images. The whole work is divided in to three parts. Each part is given name. At first level it is Noise removal with sharpening and for less percentage of noise. Proposed model discusses the method to denoise a noisy image and method of sharpening on noisy and denoised image. At this stage, in approach of increasing the visibility is given. This approach includes two steps: suppressing the noise step and the sharpening step. At second stage, it is Directional neighbourhood denoising. This method is for impulsive noise. The proposed model discusses the approach for Image Filtering. This approach comprises two steps 1) Classification of noisy and noise-free pixels.2) Removal of noisy pixels. Here noisy pixels and noise-free pixels are parted based on mean of neighbourhood pixels along four different directions in 5x5 matrix. Noisy pixels are removed by adaptive median. At third stage, proposed approach is Directional neighbourhood with wavelets. This is for Gaussian noise. The Directional neighbourhood with wavelets discusses the filtering method by multiresolution. It is combination of frequency domain and spatial domain. Input image for this model will the image which is being collected from Directional neighbourhood denoising. Wavelet transform is used forV decomposition of images and wavelet coefficient has been collected. Threshold value is calculated based on median estimation of wavelet coefficients. This threshold value is used to get new wavelet coefficient based on shrinking method. Finally inverse wavelet transform using new wavelet coefficients is processed to get the image. Assessment is done by using assessment parameter PSNR, MSE