基于SVM的多特征融合的天牛图像识别

天牛虫的图像特征提取对于天牛虫灾害的防治和监控有很重要的意义,针对目前图像识别在这个领域的应用中存在的问题,提出进行图像的特征提取和识别的方法:首先,对天牛虫图像进行二维小波变换分解,用低频图像进行特征提取,减少噪声的同时可以提高识别的准确率;然后,提取低频图像的SIFT(尺度不变特征变换)特征向量集,解决大范围的仿射失真、3维视角的改变、噪音的增加和光线的改变等造成的影响;为了提高复杂光照条件下的图像识别率,引入了颜色特征,将图像从RGB转换到HSV空间,提取图像的颜色矩作为颜色特征向量;最后,将所提取的特征作为SVM分类器的训练样本集,进而对目标图像进行识别,实验结果表明,提出的方法能够得到较好的识别效果。

SVM-based Multi-feature Fusion Longicorn Image Recognition

Feature extraction from the image of longicorn plays an important role in monitoring and preventing the pest of longi- corn. In order to solve the problem in the field of image recognition, we present a new approach to extracting the image features and i- dentifying images. First, the image of longicorn is decomposed by using the two-dimensional wavelet transform. The image features are extracted from the low-frequency image, which can reduce noise and increase the accuracy of recognition. Then the set of SIFlr feature vector was extracted, providing robust matching across a substantial range of affine distortion, changes in 3D viewpoint, addi- tion of noise, as well as changes in illumination. Color feature is added to improve the accuracy of image recognition on the condition of complex illumination, and the image is transformed from RGB to HSV space and the color matrix is extracted as the color feature vector. Finally, the extracted feature is taken as the training set of SVM classifier, thereby identifying the target image. The experiment results show that our approach is effective in the domain of image recognition.