Research On Casting Shrinkage Defect Image Automatic Simulation Algorithm

With the growing quality demand of the casting production and the use of automated testing technology, the research of casting automatic defect detection technology is becoming a hot point, also casting manufacturer is also becoming ever more demand of this technology. The development of casting automatic defect detection algorithm has been limited by the lack of defect samples. In order to provide more samples to the research of defect detection algorithm and testing it, the paper launched a study on casting shrinkage defect image simulation algorithm. Casting shrinkage is a common defect of casting defects, the study of shrinkage defect simulation algorithm can not only complement the study of defect detection algorithms, but also to develop other types of defect image simulation algorithm.In this paper, the study of shrinkage defect image simulation algorithms can be divided into the defect pattern simulation, grayscale rendering, background integration and semanticdriven simulation.Firstly, this paper analyzes the shrinkagedefect formationprinciple from the casting process, pointing out that the formationof defects are due to casting coolinglayer by layerso it has a certain randomness. From the perspective of digital image processing, the characteristics of gray, texture, shape and space is analyzed. This paper cavernous sponge shrinkage and elongated gas porosity for the study, the method of simulating shrinkage defect pattern using natural Perlin noise is proposed, using a method of digital image processing and statistical screening the defect pattern simulation which is close to real defects, and quantifying the shape features of defect pattern, a database of shape features is established. The defect pattern simulation algorithm which has a certain randomness and can be driven by parameters is achieving.Second, according to the features of x-ray image and real defects, this paper proposes a method of defect pattern gray rendering using Perlin noise parameters and gray mapping function. This paper presents the image fusion pretreatment process of irregularly shaped area, while experimental comparison fusion algorithm selection, algorithm parameter adjustment has been done, more realistic background fusion effect is achieved.Then, in order to simplify the human-computer interaction of defect simulation, using adapted Chinese segmentation algorithms, semantic parametric algorithms and semantic eventdriven process, the defect simulation method based on semantic-driven is proposed.Finally, experimental evaluation and analysis is done to demonstrate the effectiveness of the algorithm. Summed up the future development of this algorithm.