MBRSC Lab members Miss Nour Abura’ed and Miss Mina Talal presented two papers in the International Conference on Signal Processing and Information Security (ICSPIS 2018) held at University of Dubai on 7th & 8th November 2018.

Autonomous Building Detection Using Region Properties and PCA

Nour Abura’ed, Alavikunhu Panthakkan and Husameldin Mukhtar (University of Dubai, United Arab Emirates); Wathiq Mansoor (Universithy OF Dubai, United Arab Emirates); Saeed Al Mansoori (Mohammed bin Rashid Space Centre, United Arab Emirates); Hussain Al-Ahmad (University of Dubai, United Arab Emirates)
This paper proposes an algorithm for autonomous building detection in remote sensing images. The basis of the algorithm relies on the fact that each RGB channel conveys different information. Furthermore, region properties and Principal Component Analysis (PCA) are used to distinguish between buildings and other regions in order to reduce false positive cases. The images that are used to test the proposed algorithm are obtained from DubaiSat-2, which offers multispectral images with 1-m accuracy. The results of the algorithm indicate high accuracy and robustness against shadow effects. 


Detection of Water-Bodies Using Semantic Segmentation

Mina Ahmed (University of dubai, United Arab Emirates); Alavikunhu Panthakkan and Husameldin Mukhtar (University of Dubai, United Arab Emirates); Wathiq Mansoor (Universithy OF Dubai, United Arab Emirates); Saeed Al Mansoori (Mohammed bin Rashid Space Centre, United Arab Emirates); Hussain Al-Ahmad (University of Dubai, United Arab Emirates)
This paper proposes a semantic segmentation technique to automatically detect water-bodies from DubaiSat-2 images. The proposed method uses a deep convolutional neural network transfer-learning model. Several evaluation metrics such as accuracy, precision, and Jaccard coefficient are used to test our proposed algorithm. The overall accuracy for the prediction of water-bodies in DubaiSat-2 image dataset is 99.86%. 


MBRSC lab participation at ICSPIS – 2018