透射光谱成像技术及其在相似异物检测中的应用

Transmissive spectrum imaging technology and its application in detection of similar foreign materials

由于相似异物与所处背景的外观特征及反射特性极其相近，现有的光谱反射成像检测方法具有一定的局限性。本文针对相似异物采用光谱透射成像检测方法并对其中的关键技术深入研究。该方法在透射光谱成像理论模型的基础上，建立了基于综合透射差的波长优选函数，分析了近紫外-可见-近红外连续波段中异物与背景光谱透射差随波长的变化规律，筛选了区分多个异物与背景的最佳透射波长，以此构建透射光谱成像检测系统。以棉花中的白色丝状异物作为实验对象，利用最佳波长的透射成像系统获取异物特征，采用小波变换及自适应阈值的二值化处理从背景中提取异物。实验结果表明，在透射光照系数0．83条件下，该方法对相似异物的检测率达到91％，为相似异物检测提供了一种新思路。

Existing spectrum imaging detection techniques cannot effectively detect the similar foreign materials for their same appearance and reflection characteristic as the background. In this paper, a transmissive spectrum imaging method for detection of similar foreign materials was proposed and its key technology was deeply researched. According to the method, a function of optimal selection of wavelength based on synthesis transmitted discrimination between foreign materials and background was constructed after the theory model of spectral transmission imaging being developed, and the optimal wavelength differed foreign materials from background in region of near ultraviolet-visible-near infrared was selected. Then the transmissive spectrum imaging system was constructed based on optimal wavelength illumina- tion. In the actual application of detection of foreign materials in cotton, the foreign material features were acquired using optimal wavelength transmission imaging system, and the foreign materials were effectively extracted using wavelet transformation and adaptive threshold binary process. The experiment result indicated that the detection ration was 0. 91 under the condition of transmitted illumination coefficient 0. 83, and it was a novel and feasible method to detect similar foreign materials from background.