Research On Detection And Quantitative Evaluation Of Pore In Face Image

Enlarged facial pores often causes skin diseases such as acne and seborrheic dermatitis,causing people's anxiety about aesthetics.The treatment of facial pore expansion has become a common concern in the field of medical beauty,the necessity for an objective pore-analyzing system has been increased.Conventional apparatuses lack in usability requiring strong light sources and a cumbersome photograph-ing process,and they often yield unsatisfactory analysis results.Since the image analysis technology has the advantages of no contact with skin,fast speed,and reusable detection data,it has been widely used in the Medical cosmetology.In this thesis,the image analysis technology is used to detect the pores in the face image,and the quantitative evaluation method of the pores is proposed.Specific studies are as follows:1.Enhance the face image.In order to reduce the influence of local highlights on detection,this thesis adopts a face image enhancement method based on Tetrolet transform and homomorphic filtering.Compares the grayscale transformation,histogram equalization,and the single-scale Retinex image enhancement method.The contrast between the local pore area and the surrounding normal skin,the overall grayscale mean value and the grayscale variance of the face image are compared and analyzed.It can better stretch the holistic contrast of the image and enhance the clarity of the pore edges.2.Detect the area of the pores.A substep method for detecting pores in a face image is proposed.Firstly,a linear fitting-based threshold segmentation method is used to segment face image and mark the segmented target area.Secondly,the Gaussian mixture model is used to screen the markers to eliminate the interference of hair and furrows.Thirdly,the contour clusters are drawn for the selected marker points,and the actual contour of the pores is extracted based on the condition that the inscribed quadrilateral between adjacent contour lines is proportional to the corresponding side of the circumscribed quadrilateral.Comparing the segmentation method based on histogram threshold,the skin pore detection method based on HSV color space and the multi-level pore detection method,the accuracy,miss detection rate,false detection rate and average detection accuracy is used as the objective evaluation indicators.The method has better detection effect,higher accuracy and more accurate extraction of the actual contour of the pores.3.Quantitative evaluation of pores.A model for quantitative evaluation of pores was proposed.At present,the evaluation of pores is dominated by subjective scoring and grade scoring,which reduces the broad applicability of the evaluation method.In this paper,the pore size evaluation method is proposed based on the number,perimeter,area and roundness of the pores.Comparing the subjective scoring results,the feasibility and correctness of the pore evaluation method were verified.