研究支持向量机的舰船图像识别与分类技术,有效提取图像特征,提升舰船图像识别与分类效果。通过图像灰度化处理彩色舰船图像,获取灰度舰船图像;利用Gamma校正处理灰度舰船图像亮度,获取亮度适中的灰度舰船图像;利用方向梯度直方图特征提取方法,提取灰度舰船图像特征;通过局部线性嵌入算法降维处理图像特征,缩减图像识别与分类计算量;在支持向量机内输入降维后图像特征,输出舰船图像识别与分类结果。实验结果表明:该技术可有效灰度化与Gamma校正处理原始舰船图像,降低光照变化与局部阴影对特征提取的影响;该技术可有效提取船舶图像特征;在舰船图像模糊程度不同时,该技术均可精准识别与分类舰船图像,最高识别与分类误差仅有0.04。
The ship image recognition and classification technology of support vector machine is studied to extract image features effectively and improve the effect of ship image recognition and classification. Gray scale ship image is obtained by grayscale processing color ship image. The brightness of grayscale ship image is processed by Gamma correction, and the grayscale ship image with moderate brightness is obtained. The feature extraction method of directional gradient histogram is used to extract grayscale ship image features. The local linear embedding algorithm is used to reduce the dimensionality of image features and reduce the computation of image recognition and classification. The image features after dimensionality reduction are input into the support vector machine, and the ship image recognition and classification results are output. The experimental results show that the proposed method can effectively grayscale and Gamma correct the original ship image, and reduce the influence of illumination change and local shadow on feature extraction. This technique can extract image features effectively. When the fuzzy degree of the ship image is different, the technology can accurately recognize and classify the ship image, and the highest recognition and classification error is only 0.04.
2022,44(11): 156-159 收稿日期:2022-01-15
DOI:10.3404/j.issn.1672-7649.2022.11.032
分类号:TN911.73
基金项目:福建省中青年教师教育科研项目(JAT191167)
作者简介:胡牡华(1984-),女,硕士,讲师,主要从事数据科学与大数据研究
参考文献:
[1] 严明, 曹国, 夏梦. 基于水平集演化和支持向量机分类的高分辨率遥感图像自动变化检测[J]. 哈尔滨理工大学学报, 2019, 24(1): 78–84
[2] 李余兴, 李亚安, 陈晓, 等. 基于VMD和SVM的舰船辐射噪声特征提取及分类识别[J]. 国防科技大学学报, 2019, 41(1): 89–94
[3] 王成, 吴岩, 杨廷飞. 利用改进单分类支持向量机提升舰船尾流目标的检测准确率[J]. 兵工学报, 2020, 41(9): 1887–1893
[4] 付哲泉, 李尚生, 李相平, 等. 基于高效可扩展改进残差结构神经网络的舰船目标识别技术[J]. 电子与信息学报, 2020, 42(12): 3005–3012
[5] 王云艳, 罗冷坤, 王重阳. 基于流形学习的光学遥感图像分类[J]. 计算机工程与科学, 2019, 41(7): 1212–1219
[6] 王昌安, 田金文, 张强, 等. 深度学习遥感影像近岸舰船识别方法[J]. 遥感信息, 2020, 35(2): 51–58
[7] 王周春, 崔文楠, 张涛. 基于支持向量机的长波红外目标分类识别算法[J]. 红外技术, 2021, 43(2): 153–161
[8] 吕洁, 麦雄发, 谢妙. 基于二维Gabor小波和孪生支持向量机的图像识别算法[J]. 南京理工大学学报, 2022, 46(1): 113–118
[9] 周慧, 严凤龙, 褚娜, 等. 基于特征金字塔模型的高分辨率遥感图像船舶目标检测[J]. 大连海事大学学报, 2019, 45(4): 131–138
[10] 郑远攀, 李广阳, 李晔. 深度学习在图像识别中的应用研究综述[J]. 计算机工程与应用, 2019, 55(12): 20–36