The International Journal Of Engineering And Science (IJES) || Volume || 4 || Issue || 7 || Pages || PP -69-77|| 2015 || ISSN (e): 2319 – 1813 ISSN (p): 2319 – 1805 www.theijes.com The IJES Page 69 Improving Images Quality by Combination of Filtering Methods 1 D.P. Kucherov, 2 R.G. Katsalap, 3 L.V. Zbrozhek 1 Faculty of Computer Science, National aviation university, Kiev, Ukraine 2 Institute of Encyclopedic Research, National Academy of Sciences, Kiev, Ukraine 3 Faculty of Computer Science, National aviation university, Kiev, Ukraine --------------------------------------------------------ABSTRACT----------------------------------------------------------- The article deals with digital images which are corrupted by interfering signal representing random homogeneous or rare changes brightness of individual pixels. These changes matched Gaussian noise, impulse noise and their joint action. As usually to decreasing noises on the picture is used filtering. The known spatial and frequency filtering methods are leads to deteriorating sharpness images and recovering sharpness images are uses methods based on computing the second derivations, for example Laplacian. The main idea proposed approach is combined filtering methods and increasing sharpness. On the basis of quantitative criteria for estimation of quality and criteria of visual perception is studied rational combination of known types of filters. Several illustrative examples are presented that demonstrate the effectiveness of the proposed technique. Keywords - Composition, filtering, impulse noise, Gaussian noise, Laplacian, median, Wiener filtering. ------------------------------------------------------------------------------------------------------------------------------------------- Date of Submission: 24 July 2015 Date of Accepted: 05 August 2015 ------------------------------------------------------------------------------------------------------------------------------------------- I. INTRODUCTION Digital images are widely used in various fields of human activity and requirements to their quality are rising every year. Image is a natural means of communication between man and machine in any systems of processing, analysis and control. Therefore the problem of visual perception, which arose a long time, but is still relevant today, is one of the main problems in image processing [1-5]. In practice often have to deal with images are distorted by noise, which show up as imposed on initial to additional image as a mask of pixels of random colors and intensities. Visually, it looks like set granules which have different sizes, randomly are located at the image and are distorts picture's show. Especially noise is noticeable on homogeneous or dark image areas [6-8]. We will consider additive noise here. The most common types of noises are Gaussian noise and impulsive noise, and their combination [6- 13]. Gaussian noise is formed right on the images as result digitized from photographs of a portrait and natural scenes in bad conditions photographing. It may happen, for example, because of poor lighting in the photograph, as a result of the presence of electronic emission in electrical circuits and so on. In turn impulsive noise is formed in communication channels because of transferring image. For example, as a result of poor contact on transmission lines, works of powerful switching devices, as well as the action of natural phenomena such as lightning in the process of image transmission wireless. Any distortions, obstacles, signal noise worsens visual perception and image analysis, complicate their handling by different technical means. To reduce noise used filtering. Filtration images are the process that does not change the physical size of the original image, but removes certain components with desired properties. It is believed that the intensity (brightness) of each element (pixel) of the resulting image is formed from pixels intensities the some of its neighborhood. And as a term "filter" understands the system (including program), which solves the problem of filtering [9, 10]. Restoring image quality based on filtering involves the complete replacement of intensities of pixels with a sharp change of intensity compared with their neighborhood. Much work has gone into evening up this trade-off [1-12]. The approach using modified anisotropic diffusion algorithm [1] limited by application it only for images corrupted Gaussian noise. Method proposed in [2] assumes filtering monochrome image by blending the -filtering and conventional unsharp masking. In this method there is difficulty in determination optimal value  for noisy image. Image enhancement method at expense of high-pass filter is considered in [3]. Some idea in the high-resolution image [4] and wavelet [5] approach have also been proposed. Traditional and modern theoretical information about filtering can be found in works [6-8] renowned experts in the field of digital image processing. Practical software aspects solutions of problems in the field of digital image processing are described in [9, 10]. Different techniques quantify evaluation the quality of digital image processing are shown in [11-13].