A method of ear detection and extraction based on facial contour analysis

Abstract

The invention discloses a method for detecting and extracting ears in a certificate photo based on facial contour analysis, which is used for automatically judging whether the head of the certificate photo is correct, detecting the presence or absence of ears in the certificate photo, and finally extracting the ear area in the certificate photo. The algorithm analyzes two kinds of contours: the outer contour of the skin and the face contour, and is based on the observation that the contours of the human face satisfy the parabolic equation to a certain extent. Firstly, an ear detection method based on parabolic fitting of the outer contour of the skin is proposed. The outer contour of the skin is fitted with a parabola. When the fitting result is good, it means that the outer contour of the skin coincides with the contour of the face and there is no ear. Otherwise, when there are ears, a gradient- and parabolic-guided-based dynamic programming algorithm is proposed to recognize face contours. The facial contour line can segment the ear and facial skin area, so as to realize the effective extraction of the ear area. The algorithm of the invention can be used to automatically judge the eligibility of a self-portrait photo ID, and the extracted ear area can be used for post-processing of the ID photo image.

Description

Technical field [0001] The invention relates to the field of computer image processing, in particular to a method for detecting and extracting ID photos based on facial contour analysis. Background technique [0002] ID photos are one of the basic information for human identification. They are usually embedded in ID cards, passports and other document texts, or posted on application forms for admissions, employment, etc., to verify the uniqueness of user data. They are widely used in all aspects of social life. application. Traditionally, ID photos are taken by professionals. In the photo studio, the user sits in a designated position, adjusts the posture under the guidance of the photographer, and finally the photographer takes the ID photos that meet the requirements. With the rapid development of the mobile Internet, mobile phone software for ID photo selfies appeared. The user takes a face picture by himself and uploads it to the server through the network. The server autom...

AI Technical Summary

Problems solved by technology

[0004] However, none of the above methods can be used for ear detection and extraction of ID photos.

Because the ID photo is taken from the front of the face, the ear information is very little, and the internal structure of the ear cannot be seen, only the general outline of the ear can be seen, resulting in very few ear features that can be extracted; moreover, different people’s Ear shape and contour vary greatly

Therefore, the supervised ear learning and recognition method that extracts features first and then trains a statistical model is not only difficult to design, but also very complex and less robust, and is not suitable for ear detection and extraction of ID photos.

Images