Tattoo skin cross - correlation neural network

The soft biometric traits such as distinctive skin markings, tattoos based on its visual and demographic of tattoo are mostly used to identify a suspect or a victim in forensic sciences. Nevertheless, in many scenarios, police investigators type in the description of a suspect´s tattoo based on text search, therefore these classifications make it difficult to recognize associated tattoo, for example, tattoos´ criminal gangs. This paper proposed the tattoo skin recognition using cross - correlation neural network. This method has been designed into two steps: (1) the pre - processing part is keypoints feature extraction to distinguish tattoos from human skin and (2) the recognition process applies with cross-correlation neural network to classify and recognize the familiar tattoos. The tattoos database collected from prisoners in prison ministry, THAILAND. The experimental results indicated that the cross-correlation neural network outperforms the other methodologies by a wide margin. The overall accuracy obtained approximately 90.23% for tattoo skin recognition.