Emotion Recognition and Understanding for Emotional Human-Robot Interaction Systems by Luefeng Chen & Min Wu & Witold Pedrycz & Kaoru Hirota

Author:Luefeng Chen & Min Wu & Witold Pedrycz & Kaoru Hirota
Language: eng
Format: epub
ISBN: 9783030615772
Publisher: Springer International Publishing

8.2 Foreground Segmentation and Feature Extraction

Before performing the image feature extraction, it is necessary to extract the foreground containing human gesture information. For the depth foreground segmentation, we implement segmentation by detecting depth data components. As for the RGB foreground segmentation, we implement it by an iterative threshold method.

8.2.1 Foreground Segmentation Based on Depth and RGB Images

The Kinect can capture the depth data with user indexes [8]. The depth data with user index is made up of 16 binary digits, the upper 13 bits indicate the distance between the user and the device, and the lower 3 bits are the user index. The lower three bits are from 000 to 111, which respectively represent the background and user 1 to user 6. So we can easily segment the foreground image by merely retrieving the user index of the depth data.

As far as RGB image segmentation is concerned, we adopt the iterative threshold segmentation method to extract the gesture foreground [34]. It could obtain a relatively accurate segmentation with less computational overhead using only a single parameter. The segmentation algorithm works well when the histogram of the image is bimodal, and the trough is deeper and wider. In the iterative threshold segmentation, and are the maximum and the minimum gray value of the image, then the initial threshold is the average of and .

The convergence condition is that the difference of the threshold after the iteration is less than a predefined threshold, which determines the accuracy of the threshold convergence.

Download

Copyright Disclaimer:
This site does not store any files on its server. We only index and link to content provided by other sites. Please contact the content providers to delete copyright contents if any and email us, we'll remove relevant links or contents immediately.