基于机器视觉和信息融合的邻接苹果分割算

提出了利用亮度和颜色的信息融合来分割邻接苹果的方法.首先使用Lab模型对苹果图像进行分割.然后计算分割后每个区域的面积,并判断其是否为邻接苹果区域.接着在邻接区域内计算亮度信息,利用亮度产生的亮斑对邻接苹果进行分割.这样,在邻接区域以外的部分,亮度信息产生的噪声被Lab模型的信息屏蔽,而邻接区域以内的部分,具有惟一性的亮度信息可以较好分割经Lab模型处理后的邻接苹果.实验表明,此算法对邻接苹果识别非常有效,识别率大于92.89%,而且算法简单快速,平均每幅图片识别时间小于0.5 s.

Adjoined Apples Based on Machine Vision and Information Fusion

A method using intensity and color information for separating adjoined apples was proposed. First, Lab model was employed to segment the apple images, followed by each segmented area was calculated to judge whether they were adjoined area or not. Then the intensity information was added to the adjoined area, and the light spot of apple was used to separate the adjoined apples. Thus, the noise from intensity information was screened by the Lab information in the area besides adjoined apples, and the adjoined apples was segmented using the unique intensity information in the adjoined area. The result of segmentation showed that the success rate is over 92.89 %. It is useful to use the algorithm in practical recognition of apple fruits.