Title: A New Visual Vocabulary for Object Recognition

Issue Number: Vol. 8, No. 3
Year of Publication: Sep - 2018
Page Numbers: 125-128
Authors: Ismail A. El Sayad, Mohammad A. Bazzoun, Ola S. Shamieh, Samah B. El-Kashoua, Samih Abdulnabi
Journal Name: International Journal of New Computer Architectures and their Applications (IJNCAA)
- Hong Kong
DOI:  http://dx.doi.org/10.17781/P002449


Images can be represented at different levels, many approaches in the domain of image representation were developed to move from low-level to high-level image representation like Collocation patterns (moving from Visual Words to Visual Phrases), Descriptive Visual Words and Phrases (DVWs & DVPs) for Image applications and Recognition Using Visual Phrases. Moreover, many relevant articles have tackled the problem of image representation in the case of image retrieval. These articles showed advantages and disadvantages of the used methods in addressing the problem of image representation. The main purpose of this paper is to tackle these drawbacks based on the performance of image representation. What lacks in other papers is the semantic learning and not considering the spatial location of the region. In this paper, a development is proposed for a semantic coherent visual word pattern representation that considers three main points, the neighbor among visual words, frequent item set among these visual words, and the semantic learning.