基于Gabor优化选择的彩色图像边缘检测算法

Color image edge detection algorithm based on optimal Gabor wavelets

针对图像边缘检测过程中易出现图像细节信息丢失以及边缘断裂等问题,提出一种基于Gabor优化选择的图像边缘检测算法。Gabor滤波器虚部可以较好地描述像素周围的方向变化信息以及伸缩、平移、旋转不变性。利用3个尺度8个方向的Gabor虚部滤波器组提取图像边缘特征;分别计算滤波后各图像的熵值,对于同一方向、不同尺度下的滤波图像,选择熵值最大的滤波通道作为最优;根据二范数进行边缘特征融合,通过迭代式阈值分割方法获得边缘检测结果。实验结果表明,所提算法边缘定位准确且边缘细节信息丰富,边缘连续度较高。

For the problems of image details losing and edge fracture in image edge detection process,a new color edge detection algorithm based on optimal Gabor filter is proposed.The imaginary part of Gabor filter can better describe the direction change information around the pixels and the invariance of scaling,translation and rotation,and the three-scale and eight-direction Gabor filters are utilized to obtain the edge information.Calculating the entropy values of filtered images separately,for the same direction different scales filtered images,the maximum entropy value is proposed to choose the optimal Gabor filter.Fusing the filtered images in different orientations according to two norm,using iterative threshold segmentation method,the edge detection results were obtained.The experiment results show that this algorithm can obtain high edge positioning accuracy,more edge details and higher edge continuity.