Retinex Foreground Segmentation for Low Light Environments
説明
In this paper, we consider a method for improving the accuracy of the foreground segmentation based on the Gaussian mixture model (GMM) under low light environments. We utilize the GMM foreground segmentation in a system which enables fingertip gesture-input for a wearable device equipped with camera. In this system, one can operate the device with the fingertip by tapping the icons virtually projected on the space displayed through the glasses. However, in the low light environments, the number of segmentation errors of the GMM method would tend to increase due to the narrower range of change in the foreground region. In this paper, to reduce the segmentation errors, we consider applying the image enhancement based on the Retinex theory. Using an actually captured video sequence, we examine the effect of the proposed method.
収録刊行物
-
- 2018 Asia-Pacific Signal and Information Processing Association Annual Summit and Conference (APSIPA ASC)
-
2018 Asia-Pacific Signal and Information Processing Association Annual Summit and Conference (APSIPA ASC) 285-290, 2018-11-01
IEEE