Face Recognition Based On Active Shape Model Research And Implementation Of The Algorithm

Face recognition by computer models and procedures related to the establishment, through the analysis of human facial image and extract the needed information, to identify and distinguish between personal identity of a technology. The application of a wide range of face recognition in security, computer security, monitoring equipment, and other aspects of virtual reality has great use of space. Face recognition technology involves the field of artificial intelligence and pattern recognition, is a comprehensive multi-disciplinary technology. Face recognition process, including face detection, feature extraction and face recognition. States face recognition technology has been extensive research scholar, has made a wealth of achievements, many algorithms have been very mature in theory, but in practical applications, because disturbances of reality, there are still many problems. Evaluation of advanced face recognition technology is the key depends on its robustness, speed and accuracy, the present situation, can be used to achieve a real face recognition systems, also need to solve many problems.Active Shape Model is a very effective facial feature extraction algorithm, which is variable based on statistical parameters of the model. The model used for local texture directly to the shape optimization of feature points can achieve better accuracy, but, because of its gray model uses only the local texture information, the starting position of the model, the image quality is sensitive to these parameters when a large deviation occurs, it is likely to affect the overall accuracy of the results. In this regard, will present a combination of ASM with the face shape filtering algorithm, active shape model during the calibration of positioning point before the restraint used on the face shape to achieve the initial positioning of the face to reduce the burden of work and time of follow-up The experimental results show that the method can quickly human facial contours to locate, and then zoom in and out using the image to further improve active shape model the accuracy of positioning. In face recognition, human face, the color is also an important influencing factor, different people face samples, the skin color, eye color and other parts of the color differences exist. The current active shape model in the discrimination of face shape by the impact of changes in the sample position of the head larger, often resulting in errors in the results, but the normalized color histogram from the head posture changes, so, the active shape model extraction and color combination to generate the shape-color vector can not only improve the accuracy of face recognition is more important is the attitude change can be excluded due to bring the noise, the algorithm has better robust sex. Experiments show that this algorithm, face recognition, the recognition algorithm is more than practical significance.Face recognition system first to make use of samples to establish face database, the actual use of the input system to recognize the human face images, find the matching libraries face pattern, and then the next step. Using the above two improved algorithms, to achieve a set of active shape model based face recognition system, through experiments and contrast, change the algorithm is proved in two core recognition system has high recognition accuracy and robustness, You can quickly and accurately determine the location of the face, and can also eliminate the conditions to adjust the noise profile, and achieved good results.