A Novel Two-Scan Connected-Component Labeling Algorithm
This chapter proposes a novel two-scan labeling algorithm. In the first scan, all conventional two-scan labeling algorithms process image lines one by one, assigning each foreground pixel a provisional label, finding and resolving label equivalences betwe
- PDF / 1,102,902 Bytes
- 15 Pages / 439.37 x 666.142 pts Page_size
- 53 Downloads / 176 Views
Abstract This chapter proposes a novel two-scan labeling algorithm. In the first scan, all conventional two-scan labeling algorithms process image lines one by one, assigning each foreground pixel a provisional label, finding and resolving label equivalences between provisional labels. In comparison, our proposed method first scans image lines every four lines, assigns provisional labels to foreground pixels among each three lines, and finds and resolves label equivalences among those provisional labels. Then, it processes the leaving lines from top to bottom one by one, and for each line, assigns provisional labels to foreground pixels on the line, and finds and resolves label equivalences among the provisional labels and those assigned to the L. He (B) · Y. Yang · S. Li · X.Zhao College of Electrical and Information Engineering, Shaanxi University of Science and Technology, Weiyang-Qu, Daxue-Lu, Xi’an, Shaanxi 710021, China e-mail: [email protected], [email protected] Y. Yang e-mail: [email protected] S. Li e-mail: [email protected] X. Zhao e-mail: [email protected] L. He Faculty of Information Science and Technology, Aichi Profectural University, Nagakute, Aichi 480-1198, Japan Y. Chao Graduate School of Environment Management, Nagoya Sangyo University, Owariasahi, Aichi 488-8711, Japan e-mail: [email protected] K. Suzuki Department of Radiology, Division of the Biological Sciences, The University of Chicago, Chicago, IL 60637, USA e-mail: [email protected] G.-C. Yang et al. (eds.), IAENG Transactions on Engineering Technologies, Lecture Notes in Electrical Engineering 229, DOI: 10.1007/978-94-007-6190-2_34, © Springer Science+Business Media Dordrecht 2013
445
446
L. He et al.
foreground pixels on the lines immediately above and below the current line. With our method, the average number of times for checking pixels for processing a foreground pixel will decrease; thus, the efficiency of labeling can be improved. Experimental results demonstrated that our method was more efficient than conventional labelequivalence-based labeling algorithms. Keywords Computer vision · Connected component equivalence · Labeling · Pattern recognition
·
Fast algorithm
·
Label
1 Introduction Labeling of connected components in a binary image is one of the most fundamental operations in pattern analysis, pattern recognition, computer (robot) vision, and machine intelligence [6, 20]. Especially in real-time applications such as traffic-jam detection, automated surveillance, and target tracking, faster labeling algorithms are always desirable. Many algorithms have been proposed for addressing this issue, because the improvement of the efficiency of labeling is critical in many applications. For ordinary computer architectures and 2D images, there are mainly two types of labeling algorithms: (1) Raster-scan algorithms. These algorithms process an image in the raster-scan way. There are multi-scan algorithms [7, 9, 25] and two-scan algorithms [8, 10–15, 18, 19, 22]. (2) Label propagation algorithms. These
Data Loading...
