ABSTRACT
In the event of a crime, a facial composite generated from an eyewitness’ memory often constitutes the first and only means available for police forces to identify a criminal suspect. In the most general sense, a facial composite system is a tool designed to allow the expression of the facial appearance
Introduction 235 Background to Facial Composite Systems 237
Statistical Appearance Model 241 Training: The Generation of the Facial Appearance Model 242 Decomposition into Model Parameters 245 Facial Synthesis from Model Parameters 246 Relating Model Properties to Facial Appearance 246 Uses and Limitations of the Appearance Model 247 Stochastic Search Algorithm 248 Systematic Search 251
Systematic Operations in the EFIT-V System 253 Locking Facial Features 253 Feature Clone Tool 254 Blend Tool 255 Facial Attribute Manipulation 255 Training Process 255 Local Feature Manipulation 257 Managing Feature Manipulation in Appearance Space 257 Textural Effects in Composite Images 260
Application of EFIT-V 261 EFIT-V Construction Procedure 261
Effective Composite Systems and Future Development 265 Summary 267 References 268
retained in the witness’ memory in some tangible form, typically digital or a paper, hard copy. The desired outcome is that the generated composite will be of sufficient accuracy that subsequent display to members of the public or selected police officers will result in recognition, and that the name of the suspect will then be revealed. In most cases, the generated composite image will not be accurate enough to result in definitive recognition but may nonetheless provoke members of the public who recognise basic similarities to provide the names of possible suspects. Largely, it is the combination of the composite with other basic information such as age, build, domicile, and the type of crime that results in the provision of suspect names.
