Automatic Color Correction And Highlight Removal For Identification Photos

Identification(ID) photos are widely used in our daily life and appear in all kinds of certificates such as ID cards, passports and driver’s licenses. The government restricts the quality of ID photos while professional applications are difficult for ordinary people and can’t process a batch. An automatic ID photo processing system is in great need.This paper mainly describes tow functions of ID photos color optimization-facial color correction and highlight removal. Due to the light condition, devices and skills, ID photos suffer from darkness, yellowness, redness, symmetry area light inconsistent. The main works of this article are facial color correction, local area light enhancement, symmetry area light adjustment and highlight removal. Because of the particularity of ID photos, we must ensure the reality when process them. The ratio of facial color must lie in a given range; no artifacts are allowed around the boundary when local areas are edited; these problems are like a big stone lying before us.We propose a color mapping method based on quadratic function to correct ID photos’ facial color. This method can automatically process ID photos, the results satisfy the government’s request; we use blurring and classical editing propagation method to edit local area in the face and prevent this editing affecting other area in the face which is one weakness of the classical editing propagation method.Highlight often appears in the cheek, forehead in ID photos, we use Poisson method to compress the value of pixels in highlight region. We change the guidance function to make the compressing altitude of one pixel lies on the gradient norm of this pixel. We achieve our goals as well as keeping the reality of the ID photo through this strategy.The system offers some interactive tools in consideration of some poor results the automatic methods make. When the user is not satisfied with the result, he can use the interactive tools to do some tiny adjustment. The system can automatic process ID photos with 70% success rate and improves work efficiency greatly. At the same time, it offers some interactive tools when automatic processing fails.