الفهرس 
IntroductionOnly 14 pages are availabe for public view
 |  | from | 167 |  |  |
| | | from | 167 | | |
Abstract
Digital images have been increased rapidly due to the usage of users to the Internet, digital camera, and cell phones. The huge amounts of images are used for different purposes: business, education, recording/sharing events, remembering, etc. Most commercial and online shopping companies rely on images to attract consumers to visit and buy their products. Image classification is a core task for these applications. The performance of a classification model depends on two main factors: the features extracted from the images and the classification techniques. Many feature extraction algorithms are proposed by researchers such as Local Binary Pattern, Maximally Stable Extremal Regions, Speed Up Robust Features, and Deep Convolutional Neural Networks. This thesis experimentally compares among these different feature extraction algorithms by applying different classifiers with each one. So, it builds a classification model for the general digital images based on the conducted comparison. The experiment shows that using the residual neural network as a feature extraction with the support vector machine as a classifier gives the highest accuracy. The thesis also solves another interesting problem of human action recognition in still images which is a core for many computer vision applications. Despite the advances in computer vision, human action recognition in still images is a challenge as these images have no motion features like videos and the occlusions of human poses and objects in these images. This thesis proposes two different models: features-based and classification scores-based models to solve the human action classification problem. The two models are validated with the datasets: Pascal VOC and Stanford 40 actions. The results show that the features-based model outperforms the scores-based one. It gives a mean average precision of 86.55% and 84.6% on the two datasets, respectively. The experimental results and discussion are shown in Chapter 6.