Face detection Techniques

(Last Updated On: October 7, 2017)

Face detection in image processing have a big impact in research and real time applications. A big number of algorithms and techniques are presented in Matlab and OpenCV for Face Detection. Face detection is the way to detect different faces and face features. Face detection algorithm use difference and learning based techniques to identify the face from scene. Face detection is the important part of the applications for security, surveillance and automated systems. There are different models available for Face detection like Haar Cascades Face detection, feature based detection, viola jones based detection, Histogram oriented detection, segmentation based detection etc. Below are few steps performed during detection:

  • Preprocessing
  • Features extraction using HAAR and Algorithm for classification

In preprocessing starting steps are considered. First we take frames from scene. These frames contains different type of noise which is removed by applying corrections. Gamma correction are famous and best in all correction techniques and produce most clear and top results with very low error ratio. After correction color conversion is also part of preprocessing. For better processing and results frames are converted from RGB to grayscale. Gray scaling is best method to enhance speed of algorithm because gray scale is simples and light   ever format than all other formats .The frames we get are in RGB or any other format which are heavy formats and their computation cost is also high so it becomes difficult to handle and process RGB frames at real time. So frames conversion from any format to grayscale becomes important for better performance as well. These preprocessing steps are not bench mark any of these can be applied because nobody know In start which filter and what value they required or will suit the scene.

After preprocessing steps next step come face detection or face  feature extraction which is very important for detection at the frames or images we get as input contain a lot for irrelevant info which not required and take processing time and cause noise in result. Therefore it is important to extract and the main object and remove extra objects from scene. When we get frames our required face can be at any place in frame so we traverse the whole frame and check where we can get our required face or specific feature.


Write a Reply or Comment