Design And Implementation Of The X-ray Casting Detection System

Nondestructive testing technology as a new technology plays an importantrole in the actual production applications, while the X-ray detection is non-destructive testing as a representative of non-destructive testing technology,product quality control to ensure the efficiency of production plays a huge effect.Although the X-ray image acquis ition technology, the rapid progress, is no longerbased on the film presented in the form, but in digital images. However, thedetection of X-ray digital image is still largely done manually. This approach isnot only a heavy workload, and test results are vulnerable to subjective factors ofthe testing personnel, it is difficult to guarantee the detection of X-ray detectionefficiency and detection accuracy. Gradually become the dominant element in theactual production in the automation of the growing popularity of today, theefficiency and accuracy, while the traditional manual inspection shows just theopposite, cannot meet these requirements.In recent years, as image processing, rapid development and patternrecognition technology has become more sophisticated X-ray detection and thecombination of these technologies; it ushered in a new spring. Automaticdetection method based on image processing and pattern recognition of defectscan greatly improve production efficiency and the degree of automation, easy toimplement information integration to meet the requirements of the digital,automated production.For the situation described above, the paper studies X-ray casting DetectionSystem Design and Implementation. X-ray casting detection system using imageprocessing and pattern recognition technology. Adequately address the impact ofsubjective factors in the detection and identification, greatly reducing manualintervention in product testing, improve production efficiency and automation.According to the X-ray the castings detection process, the system is dividedinto three main parts to study, respectively, image pre-processing, productstandard templates training and product testing.Image pre-processing part is divided into image registration and imagebrightness adjustment of two parts. This article researches on the imageregistration algorithm, and selects one of the more simple and practical algorithmto achieve. As about the brightness adjustment, due to special circumstances inthe actual production, this article uses a relatively simple algorithm, so it is only a brief introduction in this paper.This article uses Gaussian mixture model to achieve the training of theproduct standard template, the product standard template directly affect theaccuracy of product test results. By the principle of the Gaussian mixture modeland the actual production, our aim is to find a product to create high-qualitystandard template.This article uses fuzzy pattern recognition to classify defect, defect detectionprocess is to use Gaussian mixture model to detect the foreground image defectclassification process, the paper studies fuzzy pattern recognition theory, relatedon the product testing section the collection of accurate classification of defectsdetected.