Many image segmentation algorithms have been proposed to partition an image into foreground regions of interest and background regions to be ignored. These algorithms use pixel intensities to partition the image, so it should be good practice to choose an appropriate background color as different as possible from the foreground one. In the case of a unique digitizing operation the user can make the choice of background color by himself in order to obtain a good result in the segmentation process, but in the case of several digitizing operations it would be useful to automate the whole process by removing any decision of the user about the choice of background color. Furthermore modern instruments allow capturing images with a high resolution characterized by a huge number of pixels, and pose speed problems to the image segmentation algorithms based on an idea of local thresholding. In this work an approach that adapts a widely used method for detecting moving objects from a video, called background subtraction (foreground detection), to the image segmenta- tion framework is introduced. This approach combines local and global thresholding techniques to take advantage of the computational efficiency of the former and the accuracy of the latter. It provides good results in segmentation, and allows automating the process when foreground color of images is not constant, as well as speeding it up significantly. An ap- plication to the real data concerning botanical seeds is presented in order to compare, from a statistical perspective, the results derived from the proposed approach with those provided by standard image segmentation methods.

Classification of Images Background Subtraction in Image Segmentation

MOLA, FRANCESCO;FRIGAU, LUCA;CONVERSANO, CLAUDIO
2016-01-01

Abstract

Many image segmentation algorithms have been proposed to partition an image into foreground regions of interest and background regions to be ignored. These algorithms use pixel intensities to partition the image, so it should be good practice to choose an appropriate background color as different as possible from the foreground one. In the case of a unique digitizing operation the user can make the choice of background color by himself in order to obtain a good result in the segmentation process, but in the case of several digitizing operations it would be useful to automate the whole process by removing any decision of the user about the choice of background color. Furthermore modern instruments allow capturing images with a high resolution characterized by a huge number of pixels, and pose speed problems to the image segmentation algorithms based on an idea of local thresholding. In this work an approach that adapts a widely used method for detecting moving objects from a video, called background subtraction (foreground detection), to the image segmenta- tion framework is introduced. This approach combines local and global thresholding techniques to take advantage of the computational efficiency of the former and the accuracy of the latter. It provides good results in segmentation, and allows automating the process when foreground color of images is not constant, as well as speeding it up significantly. An ap- plication to the real data concerning botanical seeds is presented in order to compare, from a statistical perspective, the results derived from the proposed approach with those provided by standard image segmentation methods.
2016
Image segmentation; Background subtraction; Fore-ground detection; Thresholding; Computational efficiency; Classification trees; Classification accuracy
File in questo prodotto:
File Dimensione Formato  
Acta_Fin.pdf

Solo gestori archivio

Descrizione: articolo pubblicato
Tipologia: versione editoriale (VoR)
Dimensione 2.27 MB
Formato Adobe PDF
2.27 MB Adobe PDF   Visualizza/Apri   Richiedi una copia

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11584/192587
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus ND
  • ???jsp.display-item.citation.isi??? ND
social impact