SUSAN算子和Hough变换在杨梅果实识别中的应用

Applications of SUSAN algorithm and Hough transform on recognition of bayberry fruit

研究自然场景下成熟杨梅的识别技术。根据杨梅果实的特点,在RGB模式下,选用R-G颜色特征用最大类间方差阈值分割法Otsu对图像进行分割去除复杂背景;再利用SUSAN算子不需要梯度计算的特点,自适应选取阀值进行边缘检测;然后通过Hough来实现杨梅果实的识别;最后通过实验结果验证本文方法的有效性。

The recognition method of bayberry fruit in natural scene was researched.According to the features of bayberry fruit,the R-G color features was selected in RGB color mode to remove the complex background of image based on the Otsu segmentation method.As the SUSAN operator does not need the gradient calculation,an adaptive threshold method of edge detection was proposed,and Hough transform was used to realize the recognition of bayberry fruit.Finally,the experimental results verified the validity of this method.