Paper Information

Journal:   JOURNAL OF FACULTY OF ENGINEERING (UNIVERSITY OF TABRIZ)   SPRING 2007 , Volume 34 , Number 1 (46) ELECTRICAL ENGINEERING; Page(s) 65 To 71.
 
Paper: 

GRASS FIELD SEGMENTATION IN FOOTBALL IMAGES

 
 
Author(s):  KANGAR LOU K., KABIR E.
 
* 
 
Abstract: 

Grass field segmentation is the first step toward player tracking, compression and content based football image retrieval. In this paper we present two pixel-based techniques applied for grass field segmentation in football images. In the first method, the hue and a new texture feature are used for segmentation. In the second method, the hue and three wavelet features representing the spatial variation of image intensity are used to classify the image pixels into the grass and non-grass sets. Considering the feature space, in both methods, the classification process is performed using an MLP classifier. In order to evaluate the performance of the proposed methods, 800 grass and non-grass samples are selected from 40 football video clips. The test results indicated that the grass and non-grass classification accuracy were 99 and 99 percent in the fist method and 100 and 99 percent in the second method, respectively.

 
Keyword(s): FOOTBALL, GRASS, SEGMENTATION, CLASSIFICATION, COLOR, TEXTURE, WAVELET TRANSFORM
 
References: 
  • ندارد
 
  Persian Abstract Yearly Visit 58
 
Latest on Blog
Enter SID Blog