融合Sobel-SIFT的花朵图像的特征提取方法

Feature Extraction Method for Flower Images Based on Sobel-SIFT Algorithm

由于花朵图像特征的复杂性,往往需要先对原有图像进行花朵轮廓检测,再进行特征点提取.本文在传统的SIFT算法的基础上,提出了改进的Sobel-SIFT算法,以适用于花朵图像的特征提取.该方法首先针对传统的Sobel算法进行了方向模板的扩展,其次对检测到的边缘特征进行细化处理,最后使用SIFT特征提取算法对花朵图像进行特征点提取.试验结果显示该算法比原有算法获取了更为准确的特征点,验证了本文方法的有效性.

Based on the traditional SIFT algorithm,an improved Sobel-SIFT algorithm was proposed to be applied to the feature extraction of flower images.Due to the complexity of the flower image features,it is often necessary to perform flower contour detection on the original image before extracting the feature points.Firstly,the method was extended to the traditional Sobel algorithm.Secondly,the detected edge features were refined.Finally,the SIFT feature extraction algorithm was used to extract the feature points of the flower image.The experimental results show that the algorithm obtains more accurate feature points than the original algorithm and eliminates some redundant feature points,which shows the effectiveness of the method.