Face Feature Detection And Analysis Used For Region Of Interest(ROI) Image Coding

ii The face feature detection and analysis used for Region of interest image coding is the important premises of ROI image coding, no other than locating and detecting the face ROI feature accurately, it can provide effective coding region for ROI image coding. At the present time there are four face feature detection algorithm: method of basing on knowledge regulation, method of feature invariant, method of template match and method of basing on appearance.The paper adopts two methods. One kind the method which unifies the feature invariant and the geometry characteristic, uses for to carry on the examination and the analysis to the color image. This method uses the YCrCb color space, illumination compensation, the similarity computation, the binary, the inflation and corrodes the operation, the hole examination and the small region eliminates, the skin color region mark, Exploiting the watershed algorithm to resolve the problem of face adjacency, then leaves the human face with the rectangular collimation marks, finally uses the geometry characteristic method to abstract the face ROI and marks with the asterisk; Another kind the method which unifies the template match and the geometry characteristic, uses for to carry on the examination and the analysis to the gradation image. This method first uses Matrox Imaging the Library function to locate the human face region of interest (eye, nose, mouth), then carries on the localization using the geometry characteristic method to the human face, uses the rectangular collimation marks to locate the eye, the nose, the mouth, the face position separately.Through the trial of color image and gradation image, it can draw the conclusion that the accuracy of face feature detection used for region of interest image coding is so high, and provide the region of interest image coding for effective region of interest.