Defect Edge Detection on CT Images of the Grain in Solid Rocket Motor

Abstract

Defects in the grain of solid rocket motor (abbreviated grain) will seriously affect the safety of remote fire strikes weapons, industrial CT testing is currently the most effective nondestructive test for solid rocket motor. Edge detection of defects on CT images of grain is a crucial step in the safety analysis of solid rocket motor. The existing edge detection algorithms for the defects in CT images of grain are mainly based on spatial domain operations, but the defect size in the CT images is not constant and many noises are included. The first derivative, second derivative edge detection operators in spatial domain operations are greatly affected by noise, which needs to be balanced between noise reduction and edge accuracy, and detection is not good for defects with smaller size. In this paper, we address these shortcomings based on the frequency domain approach. The grain images were subjected to wavelet transform to extract edge information. Experimental results show that it has high detection accuracy, and noise reduction is better, which avoids the contradiction between noise reduction effect and precision requirement.