中文摘要：

针对无明显的直线特征的仓储害虫图像,提出一种基于区域和边缘的霍特林变换旋正倾斜害虫图像的方法。首先通过自动搜索的方法确定储粮害虫的最小外接矩,利用自适应阈值对害虫图像二值化;提取最小外接矩内部害虫图像区域和边缘坐标,计算坐标数组的特征向量并构成霍特林变换矩阵;对原害虫图像区域和边缘经变换矩阵作用后实现对倾斜害虫图像的旋转矫正。试验结果表明,基于区域的霍特林变换能更快速准确地测量出倾斜害虫图像的变换矩阵并进行旋转操作,而且具有更好的抗噪性能和鲁棒性。

英文摘要：

According to the feature that images of stored-grain pests have no obvious lines,a new rectification algorithm based on Hotelling transform about the area and edge of pests was presented.Firstly,the minimum exterior rectangle was gotten through automatic searching,and image binarization was made with adaptive threshold.Then the coordinates of the region and edge of pests within the rectangle were drawn out,and eigenvectors of covariance matrix were calculated and thus the Hotelling transform matrix was constituted.The images of original slantwise pests were rectified by the matrix of Hotelling Transform.Experimental results show that the new method can rectify images of the stored-grain pests efficiently and accurately with high noise endurance and application adaptability.