An EM multi-threshold image segmentation method and device based on a histogram

The invention discloses an EM multi-threshold image segmentation method and device based on a histogram. A two-dimensional Gaussian mixture model is adopted; a two-dimensional histogram curve is fitted by using a two-dimensional GMM; an EM algorithm is adopted to obtain a distribution parameter of t...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: GONG LANFANG, GUO LING
Format: Patent
Sprache:chi ; eng
Schlagworte:
Online-Zugang:Volltext bestellen
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:The invention discloses an EM multi-threshold image segmentation method and device based on a histogram. A two-dimensional Gaussian mixture model is adopted; a two-dimensional histogram curve is fitted by using a two-dimensional GMM; an EM algorithm is adopted to obtain a distribution parameter of the GMM; on the basis of constructing a two-dimensional histogram; a two-dimensional hybrid model parameter estimation derivation method is analyzed; the number of the mixed components is determined by an adaptive method; statistical distribution of the pixel points is obtained; the image segmentation is completed by determining the class; according to the method, the distribution class of pixel points is obtained, on the basis of an oblique division method, the more the confidence coefficient isused for regional division, the longer the component is, the longer the iteration step number and the training time are, the wrong division of regions is avoided, the intercept of oblique lines of field gray division is conv