Algorithm Research On Facial Features Localization And Segmentation Based On Online Immunization

In image research and application,image recognition is one of the purposes of image processing,and image location and segmentation,measurement is the basis for image recognition.Artificial immune algorithm preserves several characteristics of biological immune system,such as good convergence,robustness,good computing performance.So the idea of immune optimization in pattern recognition,fault diagnosis,computer security and other fields have been widely practically applicated.In this paper,the immune algorithm is applied to the optimization of image location and segmentation by multi-directional selection and comparison,in order to obtain more accurate localization and segmentation results than the traditional algorithms.In this paper,the facial features localization and segmentation techniques are outlined,and the basic methods of facial features localization and segmentation are introduced in detail.The advantages and disadvantages of various segmentation methods are compared and analyzed.At the same time,the immune algorithm and the relationship with image segmentation are researched and lay the theoretical foundation for the following research.In order to improve the contrast of the image,the colorful image of the human face is processed by the gray image processing.Then,the space is covered by each gray level component according to the principle of the histogram correction on the facial image preprocessing.At the same time,the non-linear median filter is used to denoise the face image.Finally,the image is sharpened by the laplacian operator,which makes the image sharper.At the same time,considering the robustness of the detection results and the computational complexity in the detection process,we choose a multi-parameter deformable template method to detect the human face.Firstly,the parameterized surface is used to construct the multi-parameter deformable template for the salient feature of human face.Then,the corresponding energy function is constructed by this method,and the template parameter is adjusted by global optimization to minimize the energy function,Then through the multi-template and the detection of salient features detection to achieve the initial detection of the face;Finally,using the contours of the face features further accurately detect the face.Finally,to solve the problem of poor segmentation in due to improper threshold selection traditional threshold segmentation method.In this paper,the immune optimization algorithm is applied to the image segmentation and its parameters are optimized by binary coding.Combining with the two-dimensional entropy threshold principle and fast convergence of immune optimization algorithm.An adaptive entropy vaccine operator is proposed to accelerate the generation of the optimal threshold and improve the quality of segmentation to achieve the effective localization of facial features.Finally,the simulation results show the effectiveness of the proposed algorithm.