MATLAB AND ITS USE FOR PROCESSING OF THERMOGRAMS

Summary

Thermograms or better said images from thermal cameras belong to the outputs which are useful for analysis and evaluation of biological or technological systems, which have their own heat radiation. It can be used in many areas, mainly either in industry, army, architecture or and what is important in our case in medicine. This data outputs are carrying the information, which are in next phases analyzed and processed to a final or expected form. This processing can be done in more software applications, which are not only for thermograms. One of these suitable and significant applications is MATLAB. This program working on a base of matrix algorithm and with required toolbox can analyze and process the images. The main goal was to fill the problematic of processing thermograms in MATLAB and the related methodology of experiment design and its verification on a practical base. Also this methodology can be used like a guide for less or more advanced end users of MATLAB, which are working concretely with image processing toolbox.

Keywords:
thermal camera, thermovision, thermograms, image analysis, software applications, image processing

Introduction

In this work was main issue based on the proposal of methodology of processing thermograms within which it was performed a series of verification tests. These checking tests were necessary for correct and independent working of whole process. MATLAB is not free application so there was also important to buy a license for next work in this program. The main target audiences which can derive from this research are people who are working with images not only with thermograms. In case of big popularity of MATLAB was purpose of article very simple and it was, to give the readers closer look at analyzing and processing the thermograms in MATLAB environment.

Material and methods

Image processing toolbox

This toolbox is one of most commonly used toolboxes in MATLAB. It allows work with images including editing and analyzing them. Main and significant functions of toolbox are [1][2][3]:

• the spatial transformation of image
• morphological operations
• adjoining and block operations
• linear filtering and filter design
• transformation
• image analysis and improvement

Design of methodology for processing of medical thermograms with using the MATLAB

For purposes of designed methodology, there are four major points or steps which are very important in whole process of analyzing and processing of thermograms:

• A. Import of medical thermograms into the MATLAB workspace.
• B. Analysis of imported thermograms
• C. Processing of thermograms with using the functions of image toolbox.
• D. Export of analyzed and processed images into statistical and graphical process.

Description of the four main points from the methodology

A.

• read and show of the image
• check of shown image in MATLAB workspace
• writing image like a file to the disc
• content check of a newly registered file

B.

• uses of morphological opening to estimate background
• subtract the background from the original image
• determination of the threshold images
• identification of objects in the image
• inspection of a one object
• view all objects
• calculation of the surface area of each objects

C.

• histogram equalization
• segmentation, thresholding
• edge detection
• noise reduction
  • linear filter
  • median filter
  • adaptive filter

D.

• save of processed image to the disc

Examples of practical testing of designed methodology.

Background subtraction

After using I2 = I - background, we highlight a body over the background but before that we had define the background with using command:

• background = imopen (I,strel('disk',15))
• imshow(background)

Identification of objects in the image

After use of the command cc = bwconncomp (bw, 4), we found all the objects in the binary image. Using the parameter "4" was detected 39 objects. If there were some objects in contact, they were labeled as one.

Inspection of one or more objects

When viewing a single object, a sequence of objects in the numerical order as they are pictured. In this treatment were displayed objects in the following order (2, 5, 10, 15, 20, 25, 30, 35, and 39).

View all objects

In view of all objects, we used one of the ways that visualize the connected objects. Following the establishment of label matrix, we then show it as a pseudo - indexed color image.

Edge detection

This function is used to find the overall boundaries of the object. The edges of the object are manifested as a rapid brightness changes. Image should be conducted in an area in which are no other objects that do not cause the various anomalies which reduce predictive value.

Noise reduction

After applying of a salt and pepper filter we simulated, the noise, which can be caused by the same background temperature and the measured object, poor calibration or interference of TIC device certain external influences.

Conclusion

After the practical testing we have reached these conclusions. Like most suitable functions from all of tested significantly are:

• identification of objects in the image
• inspection of a one object
• calculation of the surface area of each objects
• segmentation, thresholding
• edge detection
  and
• noise reduction

Some filters, such as "average" filter or "Gaussian" filter suitable for removing noise. For example, "average" filter is useful for removing noise from images, because each pixel is set to average neighboring pixels and they are reduced due to local differences in grain size.

Other of tested functions weren’t suitable for thermograms but only in cases of processing classical images, because there were no important benefits, which would be helpful for medical sector.

According to comparison with other software applications we give a proposal to compare MATLAB with other similar programs which allows image or video processing, for example Scilab etc.

Considering usefulness and relevance of appropriate functions, we can declare, that they can bring the quality outputs from application like is the MATLAB.

Acknowledgement

This work was supported by research grant No. 26220120066 Centrum excelentnosti biomedicínskych technológií „Centre of Excellence for Biomedical Technologies“ 11/2010-10/2013

Martin Šarik

Jozef Živčák

Department of Biomedical Engineering and Measurement

Faculty of Mechanical Engineering

Technical University of Košice

Letná 9, SR-040 01 Košice

E-mail: martin.sarik@tuke.sk

jozef.zivcak@tuke.sk

Phone: +421 915 875 024