Image segmentation method based on RGB - D fusion
Yongji Yang, Lele Wang, Hua Yu, Guangnan Lu, Zhiguo Xiao · 2020
There is too little information to use in the existing image segmentation algorithms, and it can not achieve the ideal segmentation effect. Kinect depth cameras could obtain surrounding RGB image and depth image in real-time, providing a new approach for image segmentation. In this paper, based on fusion image segmentation algorithm with RGB image and the depth image by Kinect, during the time of image segmentation in the fusion of color and depth information, first of all to align the depth image and RGB image, then to segment the image in advance with significant weighted method of the depth image, finally to make four-valued logic image segmentation. Contrast with other methods, this is simpler and more effective, and could provide accurate criterion for subsequent region merging and image recognition, and it is robuster.