World Journal of Engineering Research and Technology

Nishtha Vyas*, Anand Kumar Tripathi and Anupam Vyas

ABSTRACT

Noise is major limiting factor in digital images, due to noise quality of image is degrade. According to the effect of noise and pattern of noise the noises are categorized in different types. For improving quality of image filtering techniques are used. There are various filtering technique are reported in last few years. In this presented work the impulse noise and their effect in image is measured. In addition of that the recent contributions on the image filters for impulse noise detection and correction is also investigated. After evaluation of de-noising techniques an improved impulse noise filter is proposed. The proposed image filtering technique detects and removes impulse noise from digital grey-scale images. During impulse noise the image pixels are corrupted therefore to distinguish between corrupted and uncorrupted pixels the proposed technique incorporates regression analysis. After differentiation of corrupted pixel, these pixels are processed by calculating average value of itself and two neighbouring pixels which are uncorrupted. Corrupted pixels are not properly removed in single processing therefore in iterative manner the noise is removed from image. This processing can damage the image edges therefore in order to preserve the image edges L0 smoothing is employed with each iteration. Finally for removing the remaining noise effect a median filter is applied over the image. The implementation of the desired technique is performed using MATLAB simulation environment. Result of the proposed technique measure in term of visual quality, histogram and PSNR (pick signal to noise ratio). The proposed technique demonstrates the effectiveness with respect to the previously available technique.[1,2] The proposed method is adoptable due to their less memory and time consumption.

[Full Text Article] [Download Certificate]