OBJECT DETECTION USING VIOLA-JONES ALGORITHM

Abstract

In this paper it has been described and applied method for detecting face and face parts in images using the Viola-Jones algorithm. The work is based on Computer Vision Systems, artificial intelligence that deals with the recognition of two-dimensional or three-dimensional objects. When Cascade Object Detector script is trained, multimedia content is assigned for recognition. In this work the content will be in the form of an image, where the program will have the task of recognizing the objects in the images, separating the parts of the images in the head area, and on each discovered face, separately mark the area around the eyes, nose and mouth.Algorithm for detection and recognition is based on scanning and analyzing front part of human head. Common usage of face detection and recognition can be find in biometry, photography, on autofocus option which is implemented in professional photo cameras or on smiling detectors (Keller, 2007). Marketing is also popular field where face detection and recognition can be used. For example, web cameras that are implemented in TVs, can detect every face in near area. Calculating different type of algorithms and parameters, based on sex, age, ethnicity, system can play precisely segmented television commercials and campaigns. Example of that kind of systems is OptimEyes. (Strasburger, 2013)In other words, every algorithm that has as its main goal to detect and recognize face from image, should give as a feedback information, is there any face and if answer is positive, where is its location on image. In order to achieve acceptable performances, algorithm should minimize false recognitions. These are the cases when the algorithm ignores and does not recognize the real object from the image, and vice versa, when the wrong object is recognized as real. One of the algorithms that is frequently applied in this area of research is the Viola-Jones algorithm. This algorithm is functional in real time, meaning that besides detection, it is also possible to adjust the ability to monitor faces from video material.In this paper, the problem that will be analyzed is facial image detection. Man can do this task in a very simple way, but to do the same with a computer, it is necessary to have a range of precise and accurate information, formulas, methods and techniques. In order to maximize the precision of recognizing the face of the image using the Viola-Jones algorithm, it is desirable that the objects in the images are completely face-to-face with the image-taking device, which will be shown through experiments.