الفهرس 
Chapter 1 Introduction 1Only 14 pages are availabe for public view
 |  | from | 126 |  |  |
| | | from | 126 | | |
Abstract
This thesis studies the problem of complex human activities recognition. This problem arises from the human tendency in performing their activities in not only a sequential/ simple sequence, but also in an overlapping/ complex scenario. Complex human activities could be executed in an interleaved or concurrent manner. Many research attempts addressed the problem of simple activities recognition, but little work has been done toward the recognition of complex human activities. Moreover, the existing research attempts for complex activities recognition have some limitations. First, most of proposals are learning-based and require a training dataset of complex activities, which is difficult to obtain. Moreover, they assume that activities are linearly separable and can recognize only a maximum number of two overlapping activities. In this thesis, we proposed a unifies framework for the recognition of simple and complex human activities using a combination of the discriminative features called Strong Jumping Emerging Patterns (SJEPs) and the fuzzy sets theory. The proposed approach is designed to fit the challenges of multi-label classification, nonlinear separation, and recognition of multiple overlaps of interleaved and concurrent activities. Besides the need for a training dataset of complex activities that is difficult to obtain. The proposed approach uses a training dataset of simple activities to extract two sets of SJEPs for linear and nonlinear activities. Then, the proposed Strong Jumping Emerging Patterns and Fuzzy Sets (SJEPFS)-based recognition approach is presented to recognize simple and complex activities. We evaluate our approach using two datasets (i.e. CASAS and SICHA datasets) collected from two different labs. Experimental results show the efficiency of the proposed approach in recognizing simple and complex human activities, besides the efficiency of our approach against other competing approaches with regard to recognition accuracy and time. The proposed approach achieved an overall accuracy 94.02% using CASAS dataset, and 92.04% using SICHA dataset. With regard to the recognition time, acceptable measures were achieved reaching 0.2816 seconds at CASAS dataset and 0.4105 seconds at SICHA dataset.