Vector Based Object Identification in Spherical Images
Pratiksha Benagi, Uday Kulkarni, S. M. Meena, Sunil V. Gurlahosur, T. B. Dinesh, Pradyumna Taduri, Harikrishna · Procedia Computer Science · 2020
3D virtual environment of the real world library is built using the spherical images. Nowadays identifying an object in a spherical image is a challenging task in itself, a technique is introduced to identify a particular object in a spherical image. Our case study is towards identifying a particular book in a book-shelf image which is spherical in nature. The proposed work is divided into 3 modules, the first module includes Speeded Up Robust Feature (SURF) detection and extraction. Second module incorporates application of vector based mismatch feature removal algorithm on detected SURF features. Finally the boundary box or a marker is drawn on the book detected. Proposed technique examines which format of spherical image yields better result. Experimental results show that SURF on cubic image yield better results than SURF on equirectangular image.