Regional information entropy Demons for infrared image nonrigid registration

Infrared imaging fault detection which was treated as an ideal, noncontact, and nondestructive testing method was applied to the circuit board fault detection. Nonrigid deformation was existed between the fault circuit board infrared image and the well-performance circuit board infrared image. To solve this problem, a new Demons algorithm based on regional information entropy was proposed. The new method used regional information entropy instead of image's intensity to overcome the shortcomings of traditional Demons algorithm, which was sensitive to the intensity. The inertia parameter was introduced to improve the convergence performance, which was another improvement. In inertia parameter study, the value of inertia parameter was suitable at about 0.6. The simulated study and experiment of realistic infrared image study had shown that the proposed algorithm could match the images whose intensity has difference, while the original active Demons algorithm could not. The convergence performance with the inertia parameter had been improved about twice times in experiment. (C) 2015 Elsevier GmbH. All rights reserved.