基于Gabor小波人脸图像矫正的研究

Research on face correction based on Gabor wavelet

人脸识别技术有巨大前景,但是在现实生活中,图像容易出现光照,拍摄角度,人物姿势表情的各种变化,这都会使识别的精确度减低。本课题重点研究对人脸图像的预处理中的旋转矫正。本课题主要基于彩色人脸图像,对图像进行形态学处理,进而达到人脸图像轮廓提取(人脸分割)处理,然后利用Gabor核函数滤波进行双眼定位,最后通过几何原理进行不同角度的旋转矫正,并且存储为新的图像,为下一步人脸识别打好基础。

Face recognition technology has tremendous prospects, but in real life, prone to light of the image, camera angle, posed facial expression changes, this will cause the recognition accuracy reduced. This topic focuses on face image preprocessing in the spin recovery. This topic mainly face image based on color and morphology to image processing, thus face contour extraction （face segmentation）, and eye positioning using Gabor filter kernel function, and finally through the geometry to rotate at a different angle to correct, and stored as new image, laying a good foundation for next face recognition.