Application of improved fuzzy C-means clustering in detecting human head

  • Authors:

  • He Yangming;Dai Shuguang

  • Affiliations:

  • College of Optics and Electronics, ShangHai University for Science and Technology, ShangHai, China;College of Optics and Electronics, ShangHai University for Science and Technology, ShangHai, China

  • Venue:
  • FSKD'09 Proceedings of the 6th international conference on Fuzzy systems and knowledge discovery - Volume 3
  • Year:
  • 2009

Quantified Score

Hi-index 0.01

Visualization

Abstract

Detecting human head is the common way to calculate passenger flow. Image segmentation is the first step and its effect influences image analysis greatly. Fuzzy C-Means (FCM) is often used in this aspect. Because image has large volume of data, the speed of traditional FCM limits its application in real-time situation. In order to solve this problem, this paper puts forward an improved FCM algorithm, which does not use the pixel space but histogram space. Because the data structure of histogram of every gray image is the same and the data length of histogram is 256 units, the time that every image consumes with improved FCM is very close. Further more, it makes use of the continuity of image in passenger flow statistics, and sets the original clustering center to be the actual value of previous image, which decreases the iteration times greatly and speeds up the program. The speed of Improved FCM is several hundred times faster than traditional FCM. And the result of improved FCM is nearly equal to traditional FCM. In the end of this paper, the kernel code of improved FCM is shown, and experiment proves its good effect and real-time ability.