Abstract: Aiming at the problem that high-gray-scale X-ray negative images cannot be displayed directly on low-level monitors,an adaptive enhancement algorithm for high-gray-scale X-ray negative images based on statistical decision-making was proposed.In view of the problem that 12bit high grayscale images could not be displayed,a RAW-Optical mapping algorithm was designed to quantize 12bit images to 16bit lossless display.For the problem that the gray value after mapping was too concentrated,the image was whitened and the contrast was too low,the upper and lower limits of the cropping variable were determined according to the image quantization characteristics,and the ARAW-Optical algorithm was designed to dynamically stretch the gray scale of the image to improve the image contrast.Taking advantage of the prominent feature of weld seam structure information of industrial X-ray films,an automatic exposure discrimination model based on statistical decision-making was designed by introducing the average grayscale difference and average deviation,and an adaptive grayscale correction algorithm AGRAW-Optical was designed according to the exposure degree to achieve image brightness correction and enhance image detail information.The experimental results showed that the proposed method could effectively realize the clear display of high gray-scale images,improve the visual perception of weld information,and enhance the interpretation and identification of industrial X-ray film welding slogans,welding dates and defect information.Especially in the low-illumination X-ray negative image processing,compared with the existing advanced algorithms,various indicators had been greatly improved,the information entropy was increased by 0.58%,the effective efficiency of edge information was increased by 12.75 times,the contrast was increased by 0.24%,and the image clarity index was improved nearly 10 times.

Key words: statistical decision, adaptive enhancement, automatic exposure , image processing

摘要： 针对高灰阶X射线底片图像无法直接在低位显示器直接显示的问题,提出一种基于统计决策的高灰阶X射线底片图像自适应增强算法。首先针对12bit高灰阶影像无法显示问题,设计了RAW-Optical映射算法将12bit图像量化至16bit无损显示。其次,对映射后灰度值过于集中导致图像泛白且对比度过低的问题,依据图像量化特征来确定裁剪变量上下限,设计了ARAW-Optical算法动态拉伸图像灰阶,提高图像对比度。最后,利用工业X射线底片焊缝结构信息突出的特点,引入灰度平均差与平均偏差设计了基于统计决策的自动曝光判别模型,并根据其曝光程度设计了自适应灰阶校正算法AGRAW-Optical,实现图像亮度矫正,增强图像细节信息。实验结果表明,所提方法能够有效地实现高灰阶图像的清晰显示,提高了焊缝信息的视觉观感,加强工业射线底片焊口号、焊接日期、缺陷信息判读和识别效果。特别在低照度X射线底片图像处理中,相对于现有先进算法各项指标均有较大提升,信息熵提升0.58%,边缘信息有效率提升12.75倍,对比度提升0.24%,图像清晰度指标提升近10倍。

关键词: 统计决策, 自适应增强, 自动曝光, 图像处理