---

Introduction

New methods for acquiring data have been developed with the technological advances. With this, it has been possible to obtain more precise data and, consequently produce results with greater reliability. Among the variety of acquisition methods available, those that have volume description, as CT (Computerized Tomography) and MRI stand out. The models of volumetric data (group of data that describe a solid object from a three-dimensional space) are being greatly used in a diversity of areas as a way of inspection, modeling and simulation of objects in a three-dimensional space. Applications of this model are already found in Mechanic Engineering, Geosciences, Medicine and other areas. In the area of engineering it is sometimes necessary to use industrial CT as the only non-invasive form to inspection the interior of pieces without destroying them.

In this research we present the development of a software, which can be divided in four stages:
- The implementation of the various image-processing techniques (filters);
- The implementation of segmentation algorithms;
- Calculus of physical properties (distances, areas, volumes, etc);
- Volume rendering of the original data and the segmented regions;

Picture 1 - Main window aplication (ceramic filter tomography)



Picture 2 - Main window aplication (tooth tomography)

---

Digital Image Processing

The objective of this phase was to implement basic image processing techniques (digital filters) to improve the imperfections (presence of noise and inadequate contrasts) and contours. This type of process is not easy since it depends on the type of equipment that generates the images and the characteristics of them. Therefore some techniques might show good results for a certain type of data and may not work for another type.

Techniques implemented:
- Histogram equalization.
- Threshold
- Mean filter
- Median filter
- Edge detection (Sobel and Laplace)
- Morphologic filters (dilatation, erosion, opening, closing and edges detection)

Examples:



Another system functions:

- Image histogram


- Density function throught the line

---

Segmentation

In the following section, we will present the algorithm in the form of a pseudo-code:

BTE Algorithm:
- Look for the Starting_point;
- Current Position Starting_point;
Do
- Find the position of the first edge pixel in the neighborhood;
- From this position, find the first neighbor pixel that is not an edge pixel;
- Store the position of this pixel;
- Assign the Current_position the same as this pixel;
While Current position not iqual Starting_point;

(

more about BTE Algorithm)

Examples:

---

Volume Rendering

The most used algorithms in volumetric rendering are classified according to two categories: Direct Volume Rendering and Surface Fitting (SF). The Direct Volume Rendering algorithms are characterized by the mapping of direct elements on the plan of exhibition (screen), without the necessity of using geometric primitives as intermediary representation. The Surface Fitting "find" surface primitives (polygons) defined in the data set per point that have the same value (density) to represent the region of interest. Two techniques were implemented in this work: Ray Casting Techniques and Contour Connecting.

Characteristics of Ray Casting
- Belong to the group of direct visualization
- Ideal for objects with complex structures
- Time consuming


Examples:

Ceramic filter	Screw thread	Human head

Characteristics of Contour Connecting
- Belong to the group of surface fitting
- Makes the connections between contours generally through triangles

Examples:

---

Contact Information

Author:
Master Wilfredo Blanco Figuerola 

Advisor:
Joaquim Texeira Assis, Ph.D
e-mail

Collaborators:
Claudio Antonio da Silva (Segmentation technique, BTE)
e-mail Web page

Pedro Mario Cruz (Volume Rendering, Ray Casting)
e-mail

Tools and languages:

Tecgraf - IUP Tecgraf - CD Tecgraf - IM