基于肤色检测和导向滤波的人脸美化技术

Facial Beautification Technology Based on Complexion Detection and Guided Filter

常见的人脸美化算法在美化人脸皮肤的同时,容易模糊头发、眼睛等非皮肤区域的细节性信息,导致最终美化结果视觉效果不佳。基于此,提出了一种基于肤色检测和导向滤波的人脸图像皮肤美化技术,首先对原始图像进行导向滤波处理,然后采用肤色检测算法从原始图像中检测皮肤区域和非皮肤区域,根据检测结果,将原始图像中的非皮肤区域和导向滤波后的皮肤区域进行融合处理,最后获得美化后的图像。实验结果显示,本算法美化之后的人脸皮肤自然光滑,又保留了非皮肤区域的细节信息,视觉效果更佳,且算法简单容易实现。

The common facial beautification algorithm used to beautify the skin of the face, is easy to blurred the details of the non - skin areas such as hair and eyes, which leads to the poor visual effect of the final beautification results. Based on this point, a face image beautification technology based on complexion detection and guided filter is proposed. Firstly, the original image is proceesed by guided filtering. Secondly, skin regions and non -skin regions are detected from the original image using complexion detection algorithm. According to the detection results, the non - skin area of the original image and the filtered skin area in the filtered image are fused to obtain the final beautified image. The experiments show that the beautified face skin is smooth naturally, and the details of the non - skin areas are retained, the visual effect is better, and the algorithm is simple and easy to implement.