Refined photo hair dyeing method and system

Abstract

The invention discloses a refined photo hair dyeing method and system, and the method comprises the steps: S1, carrying out the segmentation network model training of hair sample data based on a convolutional neural network; s2, performing hair region segmentation on the original picture based on the trained segmentation network model to obtain a segmentation picture mask; s3, performing Gaussianfiltering on the segmentation image mask to obtain a filtering effect image; s4, obtaining a hair area in the original picture according to the filtering effect picture, and performing color adjustment on the hair area to obtain a first dyed picture; s5, performing Alpha fusion on the original picture, the filtering effect picture and the first dyed picture to obtain a second dyed picture; s6, calculating a hair probability graph based on the original picture; and S7, performing refined hair fusion on the original picture, the second dyed picture and the hair probability graph to obtain a final dyed hair picture. According to the method, the hair probability graph is calculated, the influence of areas such as skin and clothes on hair dyeing is effectively weakened, transition is natural, and the sense of discomfort is small.

Description

technical field [0001] The invention relates to the field of photo processing, in particular to a fine photo hair dyeing method and system. Background technique [0002] Makeup and styling are an indispensable part of women's daily life, and hairstyle and hair color have a key impact on women's overall makeup image. Therefore, how to choose a hair color that suits them has gradually become a hot issue for women. Hair dyeing has become a common method for people to change their styling. Due to the uncertainty of the effect after hair dyeing, most people take a cautious attitude towards hair dyeing. Image processing technology for virtual hair coloring has begun to emerge. After the user uploads the photo, through image processing, different dyeing effects are presented to the user. [0003] In the process of realizing the function of virtual hair dyeing, the segmentation of hair area is the most basic and most important step. It mainly focuses on segmentation based on human...

AI Technical Summary

Problems solved by technology

[0004] The above-mentioned virtual hair dyeing method based on semantic segmentation of image semantic segmentation can improve the accuracy and robustness of virtual hair dyeing to a certain extent. However, it uses a deep learning model to segment the hair region of the image, and the segmentation is not accurate. The problem is that in the segmentation results, in addition to the hair area, there are also mis-segmented skin areas and background areas such as clothes

Therefore, when the segmentation of hair regions, especially hair strands, is not yet precise, this hair dyeing method has the problems of very poor transition effect between hair and skin regions, no natural transition, and obvious dividing lines

Images