A Study of Skin Visualization with Smartphones Using Deep Learning from Medical Cameras

A method is proposed to realise skin visualisation in the beauty industry using a smartphone-based consumer electronics. The smartphone is used to capture a macro image of a part of the face, while a medical camera is utilized to capture color and ultraviolet images of the same facial area. The smartphone and medical camera images are manually aligned with precision, and these images are then employed for deep learning. By utilizing a well-trained neural network, it becomes feasible to generate a UV image from the color image obtained from the smartphone. Additionally, it is also possible to convert a smartphone color image into an image that closely resembles the color image captured by a medical camera. By focusing on a specific facial region rather than the entire face, alignment becomes achievable, enabling the generation of various types of images using smartphones.

1. Introduction

Visualizing the skin is crucial in the beauty industry, and certain beauty salons possess costly skin imaging equipment. UV images are highly effective in visualizing age spots and are captured alongside color images. UV and color images of the entire face, properly aligned, can be captured. We, on the other hand, proposed a method to convert a color image into a UV image by deep learning of the images captured by a medical camera. This is achieved by training a neural network using both color and UV images obtained by enlarging partial regions of the face [1], [2].