计算机图像识别鉴定油菜种子颜色

为了更加方便准确全面地评价油菜种子籽粒颜色,研究探索了种子颜色计算机识别技术鉴定法。采用 数码照相机采集油菜种子的原始图像,根据曝光时间计算矫正系数,用Abode Photoshop CS对图像进行三原色识 别。随机提取20粒种子的RGB值,加权平均后得到RGB平均值,再将RGB平均值转化为HSB系统,获得色相 (Hue) 、亮度(Brightness)和饱和度( Saturation) ,并对种子颜色进行评价。28个不同颜色的油菜种子测定结果表 明, HSB各因素与颜色等级(CI)成极显著正相关。以H和B值对不同类别的黄籽油菜进行分类,可以明显地将肉 眼无法判别差异的纯黄籽区分出来。与肉眼鉴定的颜色等级相比,国际颜色标准RGB、HSB中的各因素均成极显 著相关关系,且精度大大提高,较为全面地反映了籽粒颜色的色相(颜色种类) 、亮度和色彩强度(饱和度)的全面 指标。

Seed color detection by computer technology in rapeseed

In order to evaluate seed color accurately,a new method of HSB color system was developed with digital camera and computer recognization technology.The seed color picture was collected by a digital camera,the RGB value was investigated in picture with Abode Photoshop software,then converted to HSB system.The hue(H) and brightness(B) of HSB system were used in evaluating color type and light reflection of seed surface.The method was tested with 28 seeds with different colors of rapeseed,and the results showed that H and B related to the color index(CI) of human eyes evalution significantly,and the seed color could be distinguished in minor difference.