Indexing both content and concept for high-dimensional multimedia data

Thesis Type: Doctorate

Institution Of The Thesis: Orta Doğu Teknik Üniversitesi, Faculty of Engineering, Department of Computer Engineering, Turkey

Approval Date: 2018


Consultant: ADNAN YAZICI


While understanding the semantic meaning of multimedia content is immediate for humans, it's far from immediate for a computer. This problem is commonly known as the semantic gap which is difference between human perception of multimedia object and extracted low-level features and it is one of the main problems in multimedia retrieval. Thus, in order to achieve better retrieval performance, low-level content features should be combined with semantic features in an efficient way. Another critical task in this domain is efficient similarity search of multimedia object in large collections. According to various studies in the literature, using query by content and concept approaches together may not only enhance performance, but also functionality of the overall system. In this study, we focus on the retrieval process of multimedia data by combining semantic information with the content of the data in order to try to solve the semantic gap problem in an efficient way. The low-level content features are extracted and mapped from high-dimesional space into low-dimensional space by using a fast dimension reduction algorithm. Thus, we have showed that our approach can reduce the retrieval problem to a spatial-indexing task and accuracy of the retrieval performed in low- dimensional space is shown to be comparable to that of the retrieval performed in the original space. High-level concept descriptors are combined with these low-level content descriptors as a new dimension and indexed together in a single structure. We also propose another index structure which uses spatial indexing method for low-level features in order to show the effectiveness of our novel approach and we proved that our study has performance enhancement in query response time of retrieving big-sized multimedia objects since it indexes content and conceptual data together for fast retrieval.