Abstract: Face detection is a very important system used by a variety of applications like face recognition systems, automatic lens adjustment, video surveillance systems etc. These systems need to primarily identify frontal face images and use them for further processing. Thus, face detection forms a primary basis for many systems. In our system, we propose an algorithm that detects faces by four modules. In the first module, we simple will detect moving objects by frame differencing technique of motion detection. In the next module, we will crop the image based on color. We will find a specific range of color of human skin and accordingly crop the image. In the third module, we will verify the symmetry of the cropped parts. A symmetry algorithm will be applied to verify the symmetry of the given part. Finally, we will detect the eye of the face by comparing the image with various eye templates. If during any of the four modules, a part of the image does not meet the threshold of the algorithm

• Publication Date: 28-Apr-2013
• DOI: 10.15224/978-981-07-6260-5-46
• Views: 0
• Downloads: 0

Abstract: Maximum Processing computation and more time consuming task has always been a limit in processing huge network intrusion data. This problem can be minimized through feature selection to condense the size of the network data involved. In this paper, we first preprocess dataset KDD 99 cup. Then we study and analysis of two decision tree algorithms (C4.5 and standard ID3) of data mining for the task of detecting intrusions and compare their relative performances. Based on this study, it can be concluded that C4.5 decision tree is the most suitable with high true positive rate (TPR) and low false positive rate (FTR) and low computation time with high accuracy.

• Publication Date: 28-Apr-2013
• DOI: 10.15224/978-981-07-6260-5-47
• Views: 0
• Downloads: 0