| In this paper, combined with the advantages of IOT 〠Artificial Intelligence and Computer-controlled, which based on the intelligent exam as the background, analyzing the face detection technology. The first module is to check the candidates’ identity in the intelligent exam system, working for the orderly conduct of the examination well.Contents of this paper: face detection technology. The paper is focus on the face recognition module in inspection record, and researching on the design for face detection in detail. Face detection technology is the key to face recognition system. In view of checking the candidates’ identity accurately, the paper wants to achieve a better and faster detection.The main idea is to collect images by camera or other equipment for processing and analysis, judging whether there is a face in the image. If detect a person’s face, need to mark it, and extract important value from image; Simultaneously, feedback the results, achieve the identity authentication based on the face feature. The effectiveness of face detection is bound to affect the performance of face recognition. After acquiring the effective face image, using the digital image pre-treatment techniques to adjust the image. Contrasting the several kinds of frequently-used face detection technologies. Then based on the previous theoretical knowledge, choosing the detection technology which based on skin color model. Testing and summary the improved method of modeling. The face detection system is applied to intelligence exam and analysis the experimental results.The system is simulated under the indoor environment, combining with the existing algorithms, proposed a detection method based on skin color model, and improved Gaussian model. The main work is as follows:(1)Image pre-processing for the acquisition, the main work is to unify the light conditions. When the brightness is enough, the color of most face regions is unified. Then, it’s better for detection the skin areas.(2)Establishing the skin model, analyzing the suitable environment of all sorts of color space. In paper choosing the YCr Cb color space and realizing the conversion of RGB space with YCr Cb space. In this paper improving the original Gaussian model and testing it. Since implement an effective binary conversion can get a skin color model with clear outline. Then, it can extend the using range of the skin model, helps to reduce the situation of mistake or missing in the follow-up project.(3)Designing the system structure, realizing the face detection step by step. Through enhancing the image by mathematical morphology can improve the quality of binary skin model with little errors.(4)Combined with prior knowledge and the template of eyes, analyzing the relationship between lips and eyes. Following, the paper design a reasonable detection algorithm for eyes and lips. Eventually, achieving the positioning and detection accurately, and will feedback the data for face recognition in the other fields.Experiments show that the system is a applied research of IOT field, which faced on checking the candidates’ identity. Designing a face detection system based on skin color model, can remove the complex background effectively and detect faces in the images. |
