自然场景下成熟苹果的图像识别研究

针对自然生长状态下的成熟苹果图像的识别问题，提出选用R-G色差模型，改进的Ostu最大类间方差法分割图像，然后用面积阈值法消除噪声，获取成熟苹果果实的目标区域。然后采用基于OpenCV的圆形Hough变换方法对多个粘连苹果果实进行分离，通过实验统计，根据苹果果实的最小外接矩形的边长设置累加器、搜索范围和圆心距。实验表明，这种方法可以取得较好的效果。

Recognition of ripe apples images in natural growth conditions

According to the recognition of the ripe apple images in natural growth conditions,R-G color space was selected,improvement of the Ostu method was used to segment images,and the area threshold method was applied to eliminate noise and obtain the target area.Then the Hough method based on OpenCV was selected to separate multiple adhesion apples.Though the experiment statistics,according to minimum circumscribed rectangular side of apple,accumulators,search scope and distance of circle were setted.experimental results show that this method can achieve good effect.