Skull face restoration method based on end-to-end convolutional neural network

• Main Authors: PAN YUELI, HU YINING, GU ZHENGYU, WANG HAO, XIE LIZHE
• Format: Patent
• Language: Chinese; English
• Published: 25.12.2020
• Subjects: CALCULATING; COMPUTING; COUNTING; IMAGE DATA PROCESSING OR GENERATION, IN GENERAL; PHYSICS

The invention provides a skull face restoration method based on an end-to-end convolutional neural network. The skull face restoration method comprises the following steps of S1, performing data preprocessing on a human skull CT scanning image and making a data set; s2, building a convolutional neural network; s3, training the convolutional neural network built in the step S2 by using a training sample, detecting a detection sample by using the model every five rounds of training, and judging a generated prediction surface type; and S4, selecting the round with the best effect in the step S4 as a final result. According to the method, the characteristics of the skull are automatically extracted by adopting a neural network method, the facial form is predicted according to the characteristics of the skull, the flexibility is high, the process is few, the consumed time is short, and the prediction accuracy is improved. According to the method, the deep convolution codec is used for learning the mutual relation