Science and Technology
publicado em 07/04/2010 às 15h48:00
   Dê o seu voto:

New algorithm facilitates pattern recognition in images

Researcher perfected the technique of pattern recognition applications, as in the analysis of gel electrophoresis

font size

Some pieces of information on this page may have been automatically translated. Makernews is not responsible for the irregularities resulting from these translations. When in doubt,      consult the original text.

Foto:Alexandre Noma/ USP
The photo on the left is the original and the right with the background replaced
  « Previous
next »  
The photo on the left is the original and the right with the background replaced

By creating an algorithm, sequence of commands passed to a computer, the researcher Alexandre Noma perfected the technique of pattern recognition applications, such as gel electrophoresis analysis, a method used by doctors for protein separation which provides medical diagnostics and development drugs.

Another application is in computer graphics programs, image segmentation, which is part of the study field of imaging informatics. Basically, the goal is to divide a given image into different regions. One example application is the extraction of the object of interest, removing the old fund to a composition with a new background image, explains researcher on the application.

In their study, part of his doctoral thesis at the Institute of Mathematics and Statistics (IME) of USP, Noma used the technique of pattern matching points. In both cases, it is to represent an image by dots, which are interconnected to form a graph, symbolizing the spatial distribution of points. The algorithm created maps the vertices of different graphs, encoding certain features of the image.

In the case of electrophoresis, each point represents a protein. The algorithm which recognizes proteins have matches among themselves and establish such links. According to Noma, the advantage of the technique is to avoid doing everything manually. The idea is to have an operational tool that minimizes user effort.

With image segmentation, the image is initially divided into several regions. Some are marked by the traces provided by the user, indicating to the program areas of interest. They are labeled as regions of the object or background regions. The goal is to propagate the labels for the rest of the image using color information and positions of the central regions.

The fact is that in the midst of numerous pieces present in the segmentation, the contours of objects that really matter are in there (the figure), says Noma, which exemplifies the image of a landscape in which contours divide a mountain of grass, sand and sea. After separation, each piece can be worked in isolation.

The researcher points out that one advantage of this segmentation technique is that it allows you to target multiple components at once, whereas most methods only work with two components (object and background).

Other applications

Noma also explored another method within the pattern matching score for other applications. In one, the researcher uses an algorithm for coloring the cartoons. The method is to trace the similar parts of different frames of the same figure. Colorization animations have to have some standardization, so (the algorithm) can recognize an object that has appeared in a painting and you can play it in the other tables, says Professor EMI Roberto Marcondes, supervisor of Noma.

The teacher explains that this standardization does not mean that the drawing must be exactly the same for all frames, there may be changes in the position of the object in focus, since it does not show new colors and other objects that were not present in the table above. Noma remember that the technique may be useful to minimize the work of the animator.

The fourth application was studied by the researcher works with forms of marriage. The technique in this case serves to image recognition, similar to what is done with image processing programs. In this application, for example, you can make handwritten digits recognition.

To create these algorithms, Noma had help from other researchers who developed these methods: Professor Alvaro Pardo, Universidad Católica del Uruguay with electrophoresis; Luis Augusto consularis currently legal analyst from the Superior Electoral Court, with the image segmentation, and Luiz Old, National Institute of Pure and Applied Mathematics, with the coloring of animations.


Source: USP
   Palavras-chave:   Algorithm    Separation of Protein    Gel Electrophoresis    Diagnosis    Alexandre Noma    USP   
  • Share this pageShare this page
  • Share this pageCorrect
  • ShareShare
  • AlertAlert
Reduced link: 
  • You are recommending this story: New algorithm facilitates pattern recognition in images
  • Fill in the following form to send your recommendation to your friend:

  • You are suggesting a correction for this story: New algorithm facilitates pattern recognition in images

Receba notícias do iSaúde no seu e-mail de acordo com os assuntos de seu interesse.
Seu nome:
Seu email:
Desejo receber um alerta com estes assuntos:
Algorithm    Separation of Protein    Gel Electrophoresis    Diagnosis    Alexandre Noma    USP   
Leave your comment
(Required fields are marked with an *)

(Your email address will never be published or shared.)

Enter the letters and numbers below and click in the button "send"

  • Twitter iSaúde
Informe Saúde printed version

Recommend the portal
Close [X]
  • You are recommending this story:
  • Fill in the following form to send your recommendation to your friend:

RSS news from the portal  iSaú
Get the newsletter of the portal  iSaú
Recommend the portal iSaú
News from  iSaú in your blog or website.
Get news on the subject of your interest.
© 2000-2011 Todos os direitos reservados.