synopsis of image restoration

Synopsis of

(Image Restoration using Type-2 Fuzzy Logic)

Under guidance of : Mrs Anita sahoo(Astt.Professor)

Submitted by Gaurav Maheshwari 0509113141

Kokil Sahai 0509113056 Kapil Kumar Gupta 0509113053

Department of Information Technology JSS Academy Of Technical Education, Noida

INTRODUCTION :

DIGITAL images are often contaminated by impulse noise during image acquisition and/or transmission due to a number of imperfections encountered in image sensors and communication channels. In most image processing applications, it is very important to remove the impulse noise from the image because the performances of subsequent image processing tasks, such as edge detection, image segmentation, object recognition,etc., are severely degraded by noise.In our project we will restore the image by removing the noise using type-2fuzzy logic technique. In type-2 fuzzy logic technique, firstly we will fuzzify the image using some rules based on sugeno based type-2 fuzzy ,then we will defuzzify and using postprocessor we get the output which will be equal to the restored value of center pixel.

Objective

Digital signals are often corrupted by noise during signal acquisition /or transmission due to a number of imperfections caused by signal sensors /or communication channels.In most signal processing operations it is of vital importance to remove noise from the signal because the performances of subsequent signal processing tasks are severely degraded by the noise.

A good noise filter is required to satisfy two conflicting criterions of

1) to suppressing the noise while at the same time

2) preserving the useful information

Unfortunately the great majority of currently available noise filters cannot simultaneously satisy the both criterions.they either suppress the noise at the cost of distorting the useful information in the signal or preserve useful information at the cost of noise suppression performance.

Hardware and Software Requirements

Operating System: Windows XP Software Tools:-MATLAB and SMART DRAW

Matlab

MATLAB is a high-level technical computing language and interactive environment for algorithm development, data visualization, data analysis, and numeric computation. Using the MATLAB product, you can solve technical computing problems faster than with traditional programming languages, such as C, C++, and Fortran.

You can use MATLAB in a wide range of applications, including signal and image processing, communications, control design, test and measurement, financial modeling and analysis, and computational biology. Add-on toolboxes (collections of special-purpose MATLAB functions, available separately) extend the MATLAB environment to solve particular classes of problems in these application areas.

MATLAB provides a number of features for documenting and sharing your work. You can integrate your MATLAB code with other languages and applications, and distribute your MATLAB algorithms and applications.

SmartDraw – It is a UML modeling tool. SmartDraw is used to create business graphics such as flowcharts, organization charts, Gantt charts, timelines, mind maps, floor plans, and other diagrams.

Developed by SmartDraw.comLatest release 2008 / September 2007

OS Windows 2000, XP and VistaType Business Graphics

License ProprietaryWebsite www.smartdraw.com

Scope of the Solution

A image filter based on type-2 fuzzy logic techniques is proposed for detail-preserving restoration of digital images corrupted by impulse noise. Theperformance of the proposed filter is evaluated for different test images corrupted at various noise densities and also compared with representative conventional as well as state-of-the-art impulse noise filters from the literature. The proposed filter exhibits superior performance over the competing operators and is capable of efficiently suppressing the noise in the image while at the same time effectively preserving thin lines, edges, texture, and other useful information within the image.

If the noise density is high (>50%) then we cannot conclude whether the noise is due to the individual pixel or due to the edge. So to get the better result we will use the adaptive method in which we will take 5*5 pixel instead of using 3*3 pixel and we will get a better result .

OVERALL DESCRIPTION

Operator Fig. 1(a) shows the general structure of the proposed impulse noise removal operator. The operator is constructed by combining a desired number of type-2 fuzzy filters, defuzzifiers, and a postprocessor. The operator processes the pixels contained in its filtering window, shown in Fig. 1(b),and outputs the restored value of the center pixel. Each filter in the structure processes a different neighborhood relationship between the center pixel of the filtering window and two neighboring pixels. Possible neighbourhood topologies are shown in Fig. 1(c). As is seen from this figure, there is a maximum of 28 possible neighborhood topologies corresponding to a filtering operator with 28 filters. However, it should be emphasized that one does not have to use all of these neighborhood topologies in practice. For most filtering applications, a filtering operator with only a few filters will yield satisfactory performance. In this case, the neighborhood topologies fed to the filters included in the structure of the operator should be chosen to be as diverse as possible to obtain the best performance. The filtering performance of the operator may further be increased as desired by including more filter processing

different neighborhood relationships. However, this will also increase the computational complexity. Hence, the choice of the number of filters needed for a particular filtering application is, in general, an application-dependent issue, which should be determined heuristically and verified experimentally.All filters employed in the structure of the operator are identical to each other and function as subfilters. However, it should be observed that the values of the internal parameters of each of the filters will be different from the other filters even though all filters have the same internal structure and the same number of internal parameters. This is because each filter is trained for its particular neighborhood individually and independently of the others during training

WORKING

Each filter accepts the center pixel and two of its appropriate neighboring pixels as input and produces an output, which is a type-1 interval fuzzy set representing the uncertainty interval (i.e., lower and upper bounds) for the restored value of the center pixel. The output fuzzy sets coming from the filters are then fed to the corresponding defuzzifier blocks. The defuzzifier defuzzifies the input fuzzy set and converts it into a single scalar value. The scalar values obtained at the outputs of the defuzzifiers represent candidates for the restored value of the center pixel of the filtering window.The candidate values are finally evaluated by the postprocessor and converted into a single output value. The output of the postprocessor is also the output of

the proposed filtering operator and represents the restored value of the center pixel of the filtering window.

B. Type-2 fuzzy Filters

Each filter employed in the structure of the proposed impulse noise removal operator is a Sugeno-type first-order type-2 interval fuzzy inference system with three inputs and one output (fuzzy set uncertainty interval). The internal structures of the filters are identical to each other.

1. INPUTS: CENTER PIXEL TWO NEIGHBOURING PIXELS

2. MEAN OF THE THREE INPUTS: (m)

3. UNCERTAIN MEAN: (u)

4. USING SOME CONSTRAINTS WE CALCULATE THE LOWER AND UPPER BOUND OF THE MEMBERSHIP FUNCTION.: (M)

5.WE CALCULATE WEIGHTED FACTOR(w)

6. WE CALCULATE OUTPUT OF RULES(R)

7.WE CALCULATE Y WHICH IS THE OUTPUT OF FILTER.

8. WE CALCULATE D WHICH IS THE OUTPUT OF DEFUZZIFIER BY TAKING Y AS INPUT.

Defuzzifier

The defuzzifier block takes the type-1 interval fuzzy set obtained at the output of the corresponding NF filter as input and converts it into a scalar value by performing centroid defuzzification.Since the input set is a type-1 interval fuzzy set.Fig. 3. Setup for training the type-2 NF filters in the structure of the proposed operator.

Yk = [Y k1 , Y k2 ] Dk = (Y k1 + Y k2)/2

9.TAKING AVERAGE OF D FOR ALL THE FILTERS WHICH WILL BE THE RESTORED VALUE OF THE CENTER PIXEL.

Postprocessor

The postprocessor generates the final output of the proposed operator. It processes the scalar values obtained at the outputs of the defuzzifiers and produces a single scalar output, which represents the output of the proposed filter. The postprocessor actually calculates the average value of the defuzzifier outputs and then suitably truncates this value to an 8-bit integer number. The input–output relationship of the postprocessor may be explained as follows.The output of the postprocessor is calculated in two steps. In

the first step, the average value of the individual type-2 NF filteroutputs is calculated:In the second step, this value is suitably truncated to an 8-bitinteger value so that the luminance value obtained at the outputof the postprocessor ranges between 0 and 255:

Overall Network Architecture

ACTIVITY DIAGRAM FOR INTRODUCTION

ACTIVITY DIAGRAM FOR DEFUZZIFIER

ACTIVITY DIAGRAM FOR POSTPROCESSOR

ACTIVITY DIAGRAM TO SHOW WORKING

Future Scope and Further enhancements

1. This project can be enhanced with following feature in future based on business needs: 2. Ability to remove at different levels in the image, this can lead to more revenue for a business.3. This Project will be in future help for image recognition also.

Bibliography

1. www.mathworks.com 2. IEEE search papers3. Stephen J chapman for Matlab4. George J. klir/ Bo yuan for fuzzy logic