Heterogeneous Facial Image Synthesis And Systems

Image is an important source for mankind to obtain the information.As a kind of image form,face sketch plays an important role in the public safety and anime entertainment.As an important branch in this field,face sketch synthesis has the important significance in public safety.Face sketch synthesis uses the model of the relationship between the face photos and the sketches via machine learning.Then we can synthesize photo to sketch and synthesis sketch to photo.In criminal cases,as a result of the limitation of environmental crime,the police cannot get the real photo of the suspect.The forensic can draw a sketch of the suspect by the describe of the witness,this sketch is the important evidence to solve the criminal case.Because of the photos and the sketches in the database are heterogeneous,if we recognize the sketch in the citizen database,the recognition rate is low,the efficiency of solving cases is reducing.So,transform the sketch into photo is the key step.At present,most of the face image synthesis methods are all based on pixel or features.This paper proposes a novel adaptive representation-based face sketch synthesis method that different regions are represented with different features.In order to divide the test photo image into different regions,we use edge detection and facial feature point detection.To combine these features for face sketch synthesis effectively,we use a Markov networks-based framework to learn the weights of multiple representations.Our experimental results showed that the proposed method outperforms existing methods which synthesize face sketches with pixel intensities or features for the whole image.In addition,the method achieves superior face recognition and image quality assessment performance in comparison to state-of-the-art methods.Finally,we set up a face photo-sketch synthesis software system.The software use the Microsoft Visual Studio 2010 development tools,with the aim of Qt as a software development framework,and use the Open CV library as the computer vision peocessing.The software has two main fuctions:face photo synthesis and face sketch synthesis.It contains the sub function: ‘open the image,‘automatic alignment',‘manual alignment',‘synthesis start',‘save the result',and ‘print the result'.It also can choose the sex and age before the synthesis starts.The software of the display area on the right is divided into test image display area,the show of the result display area and the history synthetic images display area.