摘要
针对机械手摘除苹果果袋时的图像识别问题,提出了一种通过转换颜色空间的分割算法:将RGB颜色空间的图像转换到L*a*b*颜色空间,进行自适应阈值分割、去噪、面积提取等处理后,获得果袋分割后的图像,并计算出果袋的重心坐标。实验结果表明,当果袋遮挡面积不大于果袋总面积的25%时,重心坐标水平方向误差≤10.5 mm,垂直方向误差≤8.8 mm,满足机械手进行果袋摘除的定位要求。利用该方法进行果袋图像分割可解决光照强度对图像检测影响的问题。
To the image identification problem of a manipulator picking off the apple bag, a segmentation method which takes advantage of the color space conversion was presented. First, a RGB image was converted to the L＊ a＊ b＊ color space, after making the use of the self-adaptive thresholding, removing noise and picking-up area, the apple bag im- age can be separated from the background, and the gravity center was able to be calculated. The experiments showed that when the apple bag is sheltered less than 25%, the horizon coordinate error of gravity center is less than 10.58 mm and the vertical coordinate error is less than 8.82 mm. Therefore, this processing method can satisfy the locating requirements during the manipulator picking off the apple bag. Using this method to segment apple bag images can efficiently solve the problem of illuminative influence on the image detecting,
出处
《中国农业大学学报》
CAS
CSCD
北大核心
2006年第1期87-90,共4页
Journal of China Agricultural University
