KSC 2019
Current Result Document :
ÇѱÛÁ¦¸ñ(Korean Title) |
Performance Assessment of Background Subtraction Algorithm |
¿µ¹®Á¦¸ñ(English Title) |
Performance Assessment of Background Subtraction Algorithm |
ÀúÀÚ(Author) |
Md Alamgir Hossain
Md Imtiaz Hossain
Md Delowar Hossain
Ji Hoo Chun
Eui-Nam Huh
|
¿ø¹®¼ö·Ïó(Citation) |
VOL 46 NO. 02 PP. 1049 ~ 1051 (2019. 12) |
Çѱ۳»¿ë (Korean Abstract) |
|
¿µ¹®³»¿ë (English Abstract) |
Background subtraction is a growing research area until now because of its wide area of applications. As a result, the background subtraction research is raising day by day to increase accuracy in addition to decrease complexity. The proposed assessment measures the performance of state-of-the-art approaches which is necessary for both the academy and industry. We select different types of background subtraction approach such as frame difference or median based model, statistical-based approach, cluster-based method, and sample consensus-based method. The change detection-2012 and Carnegie Mellon data set are widely used data set for the foreground-background classification which we use for the performance measurement of the algorithm. The change detection-2012 includes six categories while the Carnegie Mellon data set contains one video of 500 frames. We mainly exploit F1-score in addition to other confusion metrics which are more acceptable metrics to calculate the accuracy of the background foreground segmentation algorithm. Experimental results are tabulated in method performance section 4.
|
Å°¿öµå(Keyword) |
Performance Assessment
Background Subtraction
Foreground Detection
Moving Object Detection
Segmentation.
|
ÆÄÀÏ÷ºÎ |
PDF ´Ù¿î·Îµå
|