Human face recognizing method based on geometric matching and split-merge algorithm

Abstract

The invention relates to a human face recognizing method based on geometric matching and a split-merge algorithm. The method comprises the following specific steps: a. cutting an image in human face positioning; b. carrying out the split-merge algorithm after the image is cut; c. conducting a grouping algorithm; d. calculating shape attributes, including (1) the position of a feature block, (2) the orientation of the feature block, and (3) an external rectangle of the feature block; e. preprocessing the image, including (1) the enhancing to image contrast, and (2) binaryzation; and f. conducting a geometric matching algorithm. According to the human face recognizing method based on the geometric matching and the split-merge algorithm, with the adoption of the split-merge algorithm and the geometric matching algorithm, the positioning to a human face is realized, the algorithm is clear and simple, a certain noise resistance ability is provided, and method accords with the characteristic that a human eye has multiple scales of vision.

Description

technical field [0001] The invention relates to the field of face recognition, in particular to a face recognition method based on geometric matching and split-combine algorithm. Background technique [0002] The location of the so-called human face is to mark the position of the human face in the photo (static image) or video (dynamic image), and select the human face. Face recognition is to compare the selected faces with the existing faces in the database to find matching files. Some literature refers to face positioning and recognition collectively as face recognition, and positioning and recognition are two main steps. A complete face recognition system involves determining whether there are faces in a photo or video, and counting, locating, sizing, identifying individuals from a database, and possibly expressions, and describing them from images of faces ( Melon seed face, red phoenix eyes, etc. are examples of "description" in daily life), or conversely select match...

AI Technical Summary

Problems solved by technology

The first type of method uses obvious prior knowledge, so the method is simple and clear, the execution speed is fast, and it has certain adaptability to the direction and expression of the face (the relative geometric relationship of facial features changes very little within a certain range of variation. ), but the accuracy rate is often not high (missed and misjudged), and it has high requirements for preprocessing, relying on all facial features to be completely extracted, so for side faces with large corners, the illumination is extremely uneven, and partial faces Being shaded (glasses, scarves, etc.) is not well adapted

The second type of method utilizes more image information, has a high accuracy rate, and is not easy to be deceived; the disadvantage is that the amount of calculation is large, and the face template used is affected by the existing data in the face database, which may have poor versatility problems (for example, face templates of different races cannot be used universally)

Images