类球形水果表皮颜色变化校正方法研究

Correction Algorithm of Lighting Non-uniformity on Spherical Fruit

摘要:

针对类球形水果表面较大的曲率变化会引起表面亮度不均，从而导致水果颜色分级评价中存在误差大、准确率低等问题，提出了二维B样条水果表面亮度不均校正算法。利用该算法分别对原始RGB图像各单通道图像进行亮度校正，然后将校正后的RGB图像转换成HIS颜色空间图像，提取色调H和亮度I分量，通过对比校正前后H和I分量图像像素灰度标准差评价校正效果。对160幅橙图像处理结果表明，校正后的图像在色调和亮度上比原始图像更加均匀，色调H分量和亮度I分量的平均标准差分别仅为原始图像标准差的21.57%和33.94%，色调和亮度均匀性得到了明显的改善。

Abstract:

In terms of spherical fruits, the lighting was usually uneven distribution on the surface of fruits due to the larger change of curvature from fruit surface. Therefore, some drawbacks such as large error and low accuracy were still existed in the grading and assessing for fruit peel color. In order to solve this problem, B-spline lighting correction method was proposed in this study. Using the proposed algorithm, R, G and B channel images abstracted from original RGB image were firstly corrected respectively. Then, the corrected RGB image was changed into HIS color space image and hue H and illumination I component images were abstracted. Finally, the correction performance was assessed by computing the standard deviation of pixels in H and I component images before and after correction. For the investigated 160 orange sample images, the result showed that the corrected images were more uniform in terms of hue and illumination. Only 21.57% and 33.94% of mean standard deviations of original hue and illumination component were obtained. The uniformity of hue and illumination was effectively improved.