Date of Award
Doctor of Philosophy in Electrical Engineering - (Ph.D.)
Electrical and Computer Engineering
Ali N. Akansu
Wayne Hendrix Wolf
Richard A. Haddad
Yun Q. Shi
This thesis is a comprehensive study of object-based image and video retrieval, specifically for car and human detection and activity recognition purposes. The thesis focuses on the problem of connecting low level features to high level semantics by developing relational object and activity presentations. With the rapid growth of multimedia information in forms of digital image and video libraries, there is an increasing need for intelligent database management tools. The traditional text based query systems based on manual annotation process are impractical for today's large libraries requiring an efficient information retrieval system. For this purpose, a hierarchical information retrieval system is proposed where shape, color and motion characteristics of objects of interest are captured in compressed and uncompressed domains. The proposed retrieval method provides object detection and activity recognition at different resolution levels from low complexity to low false rates.
The thesis first examines extraction of low level features from images and videos using intensity, color and motion of pixels and blocks. Local consistency based on these features and geometrical characteristics of the regions is used to group object parts. The problem of managing the segmentation process is solved by a new approach that uses object based knowledge in order to group the regions according to a global consistency. A new model-based segmentation algorithm is introduced that uses a feedback from relational representation of the object. The selected unary and binary attributes are further extended for application specific algorithms. Object detection is achieved by matching the relational graphs of objects with the reference model. The major advantages of the algorithm can be summarized as improving the object extraction by reducing the dependence on the low level segmentation process and combining the boundary and region properties.
The thesis then addresses the problem of object detection and activity recognition in compressed domain in order to reduce computational complexity. New algorithms for object detection and activity recognition in JPEG images and MPEG videos are developed. It is shown that significant information can be obtained from the compressed domain in order to connect to high level semantics. Since our aim is to retrieve information from images and videos compressed using standard algorithms such as JPEG and MPEG, our approach differentiates from previous compressed domain object detection techniques where the compression algorithms are governed by characteristics of object of interest to be retrieved. An algorithm is developed using the principal component analysis of MPEG motion vectors to detect the human activities; namely, walking, running, and kicking. Object detection in JPEG compressed still images and MPEG I frames is achieved by using DC-DCT coefficients of the luminance and chrominance values in the graph based object detection algorithm. The thesis finally addresses the problem of object detection in lower resolution and monochrome images. Specifically, it is demonstrated that the structural information of human silhouettes can be captured from AC-DCT coefficients.
Ozer, Ibrahim Burak, "Object detection and activity recognition in digital image and video libraries" (2000). Dissertations. 456.