Image processing has a key role in ITS system and Transportation Infrastructure management. On the other hand, system modeling is an essential step in decision-making process in order to explain, predict and control a system. Better tools for analyzing the data lead to better modeling of the system and hence, better solutions for the decision problems. Amongst the various system modeling techniques, fuzzy system modeling (FSM) provides the decision maker with valuable knowledge (linguistic variables).

In this project an existing Type-II fuzzy image processing expert system based on Type-II fuzzy function to recognize patterns in images. This approach has four main steps:

- Pre-processing
- Segmentation
- Feature extraction
- Approximate reasoning

We improve this approach by using Type-II fuzzy function instead of the traditional rule base in approximate reasoning steps. Moreover, I developed and optimized a program code in MATLAB with some computer techniques. The result code was 6 times faster than the original version. The original code was for image processing for brain tumors. However, I generalized the code and it is applicable in different fields especially in transportation systems.

In addition, I designed a GUI in Microsoft Visual Studio for this code by using MATLAB MCR. This software can be used for traffic image processing, railway and pavement monitoring system.

The following strategy is used in Approximate reasoning step, to identify the Type-II fuzzy function and to obtain the diagnosis:

Step 1- Defining membership functions

Step 1-1: Clustering the output data

Step 1-2: Tuning the parameters of membership functions

Step 1-3- Defining the suitable Type-II membership functions

Step 2- Selecting a method for function identification

Step 3- Normalizing membership functions

Step 4- Calculating augmented input matrix

Step 5- Calculating interactive cluster

Step 6- Deducing the results

The result was an article which was presented last summer in Madrid. You can find additional technical information on it:

M.H. Fazel Zarandi, M. Zarinbal, **A. Zarinbal**. (2010) “Using Type-2 Fuzzy Function for Image Processing Approaches”,IEEEWorld Congress on Computational Intelligence*.*

The original work is described well in this article on Elsevior: