Research On Facial Skin Defect Detection Algorithm Based On Image Processing

Facial skin defects include spots,acne,wrinkles,and so on.Most people have facial defects in the short or long term.The facial skin condition not only reflects the health level inside the human body,but also is an important reflection of the external image and the spiritual outlook.How to evaluate and diagnose facial skin,and how to judge the efficacy of skin care products need objective quantitative evaluation of facial skin.The traditional facial skin evaluation methods include direct observation method and professional instrument detection method.Direct observation is a subjective method with obvious shortcomings,such as inaccuracy,insensitivity and heavy dependence on personal experience.Although the accuracy of professional instrument detection is high,the operation of this method is complex,and the method is expensive and limited by the location of use.With the continuous expansion of computer image processing technology to various application fields,the detection of facial skin defects by computer image processing technology has become a new method in this field.Facial skin defect detection algorithm is the core of the facial skin evaluation system.Therefore,this paper studies the algorithm of facial skin defect detection based on image processing,which can detect and extract facial spots,acne and wrinkles respectively.The research contents mainly contain the following parts:(1)Preprocessing face image.Face detection and contour detection are used to extract facial skin region.In order to avoid the skin color difference in different positions of face,the facial skin image is divided into several non-overlapping sub-images.In order to improve the detection accuracy of the algorithm,image enhancement techniques are used to pre-process the sub-image.(2)Detecting and extracting facial color spots and acne.The color spots and acne have dominant characteristics relative to the normal skin in the corresponding color space,and have Poisson distribution characteristics in the facial region.The Cr-AngleA-H color space is constructed and transformed into a new distance space.The histogram is fitted by Poisson distribution,and the appropriate segmentation threshold is calculated.Finally,the spot region is detected and extracted.(3)Detecting and extracting facial wrinkles.Wrinkles have unique orientation and texture morphological features in facial region.The Gabor filters in different directions are designed to extract features,the maximum filter response image and texture direction field are got from those Gabor filters.Gaussian Mixture Model and Markov Random Field(GMM-MRF)are used to segment the maximum filtered response image.The segmentation result of the maximum filter response image and the texture orientation field are fused.Finally,the wrinkle region is detected and extracted.The algorithm can detect the facial skin defect objectively and quantitatively in experiment,so that users can choose the appropriate skin care products and treatment schemes according to the current skin condition.Therefore,this study has important theoretical research significance and application value.