Fluorescent Magnetic Particle For Crack Testing Research And Implementation On Image Feature Extraction Method

Fluorescent magnetic particle testing is a common method used for the surface of magnetic materials in iron or near surface defect. It has such advantages as not damaging the raw material, high detection sensitivity, and fast speed, simple and reliable process.The methods involve the crack image testing which is a subject with research and practical values booming in recent years. However, the current fluorescent magnetic particle testing system is mainly in the manual and semi-automatic form. And it’s efficiency still needs to be further improved; the general magnetic particle testing by images depends on the software on PC, thus it can not be operated due to its poor portability. This paper shall research on the extraction methods of fluorescent magnetic particle crack image features to lay a solid foundation of automatic testing.it also develops applied platform based on ARM to make a preparation for the development of portable products.The extraction of fluorescent magnetic particle crack image features in this paper has two steps: identify the magnetic particle crack and calculate the position and size of the crack. Firstly, in the process of identification, smoothing algorithm of weighted average filter is employed to have a filtering of the magnetic particle image. Based on fractal, it adopts the box dimension to make out the fractal dimension of the Fluorescent magnetic particle cracks. It also needs to make an account of the fractal dimension of positive and ne gative samples and fit the characteristic curve of fractal dimension of positive and negative samples. Choosing the threshold of based on the curve fractal dimension toidentify if the work piece has cracks. Secondly, the measurement of position and size is based on former crack testing and it adopts the region segmentation to further deal with the original image of the cracking work piece, getting the six geometric features: area, perimeter, circularity, and rectangularity, eccentricity and length; calculating the position and size of the cracking area based on the area and length to pick out the real cracking areas.Based on the above research, the writer of the paper employs an applied platform and appliesfeature extraction of the fluorescent magnetic part icle crack images into the embedded system. The key work includes the module implementation of fluorescent magnetic particle crack images. Fluorescent magnetic particle crack image extraction module can realize the fractal dimension extraction and screen the work piece with defects and cracks. Based on this information we can determine the position and size. The experiment shows that the future extraction methods can be operated stably and reliably on the embedded platform and at the same time it can also m eet the requirements of practical testing.