Зарегистрироваться
Восстановить пароль
FAQ по входу

Meijster A. Efficient Sequential and Parallel Algorithms for Morphological Image Processing

  • Файл формата pdf
  • размером 851,20 КБ
  • Добавлен пользователем
  • Отредактирован
Meijster A. Efficient Sequential and Parallel Algorithms for Morphological Image Processing
Dissertation. — Rijksuniversiteit Groningen, 2004. — 162 p.
In this thesis the focus is on efficient sequential and parallel algorithms for morphological image processing. Mathematical Morphology is a field of non-linear image processing, based on minimum and maximum operations. Good introductions to morphological image processing are [38, 86]. The aim of this type of image processing is to extract or enhance features from images based on shape. In this context, accuracy plays a crucial role, in contrast with picture processing where accuracy is usually traded for visually appealing results.
Many morphological operators can be computed very fast. However, most operators are hardly useful on their own. They are applied in sequence, adding up processing time. Most operators can be extended easily to 3D-image processing, requiring even more processing time. Morphological image processing based systems are typically used for real-time surveillance tasks in industrial systems, medical image processing, optical character recognition, texture analysis, etc.
Many algorithms for various morphological image operators have been published. The aim of this thesis is to speed up some of these algorithms by means of improvements in the original algorithms, or by the use of parallel computing techniques. Nowadays, many desktop workstations contain multiple CPUs, which could be exploited by parallelization of existing algorithms for morphological operators. Since these machines become more common, the focus is on parallelization for shared memory or SMP architectures, which can be desktop workstations with multiple CPUs as well as massively parallel shared memory supercomputers.
Introduction
A General Algorithm for Computing Distance Transforms in Linear Time
Concurrent Determination of Connected Components
A Comparison of Algorithms for Connected Set Openings and Closings
A Concurrent Algorithm for Connected Set Filtering, and its Application to Interactive Visualization
TheWatershed Transform: Definitions, Algorithms and Parallelization Strategies
Concluding Remarks
  • Чтобы скачать этот файл зарегистрируйтесь и/или войдите на сайт используя форму сверху.
  • Регистрация