Real-Time Object Segmentation from Network Camera using Touch Screens
Touch screens interfaces have become the quicker, more intuitive way to interact with surround technologies. We present an interactive object region segmentation technique that leverages the touch screen technology to the detection and identification of objects in images captured in real environments.
Our algorithmic work addresses one of the most challenging tasks in image processing and computer vision research fields, specifically, the segmentation of objects that have non-homogeneous body textures. The proposed segmentation method employs Seeded Region Growing (SRG) segmentation algorithm to extract the precise and accurate object region from other surrounding objects and backgrounds. The human input via touch screen is used to select the seed point on the object of interest within the video. Our algorithm uses the seed points as the initialization for the seeded region growing technique. The proposed system is evaluated by observing its capability to correctly segment the selected objects, while simultaneously performing invariant to the users choice of the object.
Publications in this field
- Fatema Albalooshi and Vijayan K. Asari, "Textural discrimination in unconstrained environment," IST/SPIE International Conference on Electronic Imaging: Imaging and Multimedia Analytics in a Web and Mobile World 2014, San Francisco, CA, USA, 2 - 6 February 2014