基于高光谱技术检测苹果外观缺陷

目的 利用高光谱技术检测苹果外观缺陷, 分析主成分分析法和波段比率算法研究高光谱图像的可行性。方法 在400~1100 nm波长范围内获取苹果表面的高光谱图像信息, 用主成分分析法处理高光谱下采集的苹果图像, 选取第三主成分图像进行分析, 作为最后的判别依据。波段比率算法中选取了717 nm和530 nm两个有效波段,将两个波段的图像进行比值运算。717 nm波段的图像进行阈值运算、中值滤波及形态学分析得到二值化掩膜图像, 再与二值化后的比率图像进行布尔运算, 提取缺陷的有效信息。结果 基于主成分分析法, 检测苹果表面缺陷的分级准确率为81.25%, 波段比率算法对苹果表面缺陷的分级准确率为93.75%。结论 利用高光谱成像技术下波段比率算法相对于主成分分成法更适合于实时、在线、快速检测。

Detection of defects in apples based on hyperspectral imaging technology

Objective To detect the defects in apples by hyperspectral technology, and analyze the application of principal component analysis and band ratio algorithm for detection of bruises in apple surface. Methods The hyperspectral image in the range of 400~1100 nm was acquired from the surface of apple sample. Third principal component image was selected for analysis in the research. Ratio of two effective bands of 717 nm and 530 nm were calculated for band ratio algorithm and the two band images were performed to ratio transformation. Next threshold segmentation, median filtering and morphological analysis were carried out at 717 nm band image to build binary mask image. Finally, the mask image was applied to the binarization of ratio image to extract the effective information of defects. Results The grading accuracy rate was 81.25% based on principal component analysis. Similarly, detection accuracy of 93.75% was observed based on the band ratio algorithm. Conclusion The band ratio algorithm is more suitable for application of hyperspectral imaging technology for real-time, on-line, and rapid detection.