Research On The Generation Algorithm Of The Shrinkage Cavity Defect Image

Because there are many advantages of casting technics, which is widely applied to various fields of industrial production, especially in automotive casting industry. Countries, manufacturers and users pay more and more attention to environmental protection, safety requirements and performance experience, so automati ve defects detection of castings is indeed necessary, making the algorithm of automatic defect detection is a research hotspot nomatter in domestic or foreign currently. However, the lack of a large number of defects in reality has been the obstacle of automati ve defect detection technology. This paper explores and studies the algorithm of casting shrinkage cavity defect image generation, in order to provide more training samples to the automatic detection algorithm. The casting shrinkage cavity defect is one of the common type of casting defects, the research of the shrinkage cavity defect generation algorithm not only provides auxiliary function for this type of defect detection algorithm, but also be a reference to other types of defect generation algorithm.In this paper, the automative generation algorithm research of shrinkage cavity defect image is mainly divided into three stages, including the generation stage of the original simulation defect image, the transform processing stage of original simulation defect image, the stage of background fusion.Firstly, this paper introduces the formation mechanism and classification of the shrinkage cavity defect, expounding the generation of shrinkage cavity defect due to liquid contraction and solidification contraction. From the perspective of digital image processing. After analysing the gray feature and texture features and spatial relationship features of real shrinkage cavity defect image, the features of digital image shrinkage cavity defect are obtained. Then generate random altitude data according to the improved Diamond-Square algorithm in this paper, and generate the original simulation shrinkage cavity defect image based on gray mapping. At this time the simulation shrinkage cavity defect and background image can not be directly fused, otherwise it will lead to a unsatisfactory fusion effect. Therefore, it is necessary to apply transform process to the shrinkage cavity defect image of original generation simulation.Secondly, according to the fusion method proposed in this paper, making the simulation shrinkage cavity defect image after transformation and X-ray image of real shrinkage cavity defect image fused, generating the fused image.Finally, in order to carry out the analysis and evaluation of the effectiveness of the fusion effect, this paper adopts the casting defect detection algorithm based on deep-learning to detecte defects of the final generated fused image. Experimental results verify the generated simulation defect s can be detected correctly, and be able to identify as the shrinkage cavity defects. It is proved that the three parts algorithm in the generation stage of the original simulation defect image and the transform processing stage of original simulation defect image and the stage of background fusion original stage are effective. At last summarizing the shortcomings of the algorithm and pointing out the future development direction of the algorithm.