Airport Pavement Distress Image Classification Using Moment Invariant Neural Network

ABSTRACT: In this paper we present moment invariant and neural network to classify airport pavement distress images. The moment invariant has been extracting images feature that published two-dimensional pattern recognition application method in 1962. The image pattern can be reduced to number of values such that they are description of the translation, scale, and rotation of an object in the image. In this paper we investigate the technique of moment invariant for the crack image of feature extraction. Then the neural network that we use back-propagation learning in its training, classifying these feature, and attempts to produce the desired output. The real crack images are always not pure distress features. However, the back-propagation neural network may be used to provide fitness against noise. The results indicate that using moment invariant and neural network to provide accurate classification processing in various types of airport pavement distress.