Nowadays, image processing is among rapidly growing technologies. Deblurring of an image nowadays is very important in the field of photography and multimedia, which is dependent upon various parameters, such as Gaussian noise, Shaking of the camera, incorrect focus, object motion, etc. The signal processing deals with the transmission, reception, and storing of the data. The data is in the form of analog or digital data ( Bae and Durand, 2007). In this method, filtering plays a very crucial role in the transmission and reception or reconstruction of the image. The signal involves the transmission of an image, voice, or sound. Image processing is a method to perform an operation on the image to extract some information or to enhance the image ( Shen et al., 2012). The basic purpose of image processing is observing the object to be processed, image pattern recognition, and image restoration. Blurring creates distortion in the image which makes any image unclear. Some of the examples of blur effects are shown in Figure 1 ( Dabov et al., 2008). These blurs are due to aperture of the camera and is due to the movement of the object.

Figure 1. Blur Detection in Images

Image Deblurring is of two types based on the image and operator that are known or unknown, such as blind deblurring and Non-blind deblurring. In a blind deblurring method as the name indicates the image or operator, in this case, may be partially known or unknown completely, whereas in Non blind deblurring in which image, as well as operator are known ( Zhou et al., 2011). Blur image can be of various types, such as motion blur, lens blur, turbulence blur, and postprocessing blur. An image is a two-dimensional representation of the data on a surface, which consists of some information. It may be colored image or black and white image. The blur is that the objects in the visual image are not visible ( Whyte et al., 2012). There are various reasons of Blurring of an image, such as due to the movement of the camera during capturing of an image, due to out of focus optics, due to the turbulence of atmosphere, short exposure time, and the confocal microscopy suffered from scattered light distortion. The image processing via MATLAB provides a platform to observe the blur and restore the image to its original form ( Tang et al., 2013).

In this paper, a method has been proposed to detect the blur level in the image and restore the image to its original form, i.e. deblurring of the image in such way that the image can be clearly visualized.

There are various methods already proposed to deblur any blur images. Levin et al. (2007) proposed a method to restore the image captured from the modified camera. This camera captures a single refocus blurred image, which is recovered. In this method, inorder to improve the accuracy of the blurred image, sparse prior along with coded aperture is done. Zhou et al. (2011) proposed a method in which to capture an image, it consists a pair of optimized coded apertures so that a high quality in-focus image can be obtained from the two images. It is important for any method to get some information of the blurred image to be recovered. Therefore, an accurate image can be obtained. Oliveira et al. (2014) proposed a Randon-C transform method to obtain the actual image. Cheong et al. (2015) proposed a method, to reproduce the original image by considering the blur kernel to be a Gaussian function. This is done by obtaining the amount of blur from the ratio of local variances of the first and second-order derivatives. Chan and Ngugen (2011) proposed a local blur estimation method to speed up the deconvolution method and to control the blur change which occurs abruptly at the depth of discontinuities.

In this method, a criterion to adjust the regularization parameter and the criterion for stopping the operation have been proposed. The methods, such as blind image deblurring and Non blind deblurring, have been used in this paper. This is done by assuming that white noise is present within the space provided of the image. By implementing this method based on the measures of the spectral whiteness, the fitness of the current estimation can observed. The fitness of the disturbed model is to be assessed initially ( Mehra, 1971). The residual measuring is one of the frequently used methods to assess the accuracy of any model. But the whiteness measuring residual criteria has not been used in the deblurring or deconvolution of the image. The main focus of this paper is on Blind Image Deblurring (BID), which is performed by choosing regularization parameter and based on the stopping iteration fitness of the image. The flow chart of this methodology is shown in Figure 2.

Figure 2. Flow Chart for Deblurring of Image

In this paper, a method called whiteness measurement of the residual image was proposed. This deblurring is based on blind and nonblind estimation problems. This method is applicable for color as well as monochrome images for both blind and nonblind image deblurring methods.

Expression of giving thanks are just a part of those feelings which are too large for words, but shall remain as memories of wonderful people with whom I have got the pleasure of working during the completion of this work. I am grateful to RITEE, Raipur which helped me to complete my work by giving encouraging environment. I would like to express my deep and sincere gratitude to my supervisor, Assistant Professor Dr. Sanjivani Shantaiya. Her wide knowledge and his logical way of thinking have been of great value to me. His/her understanding, encouraging and personal guidance have provided a good basis for the present work.