摘要
红外序列图像的准确分割是自动目标识别的关键,而当图像背景复杂时,传统的图像分割技术往往难以满足要求,为此,提出了基于支持向量机的红外序列图像分割方法。序列图像中的部分帧被作为训练样本,通过选择适合的模型参数,运用支持向量机方法建立学习机器,将后续图像帧中的目标从复杂的背景中识别出来,从而实现红外图像分割。实际红外序列图像分割表明,基于支持向量机的图像分割方法不需要复杂的预处理和后处理工作,分割效果理想,对于小目标的图像,识别正确率可达 99%。
Accurate segmentation of infrared image sequence is the key for automatic object recognition (ART). However, the traditional methods of image segmentation are always unable to perform well under the condition of complex background. For this reason a method for infrared image sequence segmentation based on support vector machine (SVM) is proposed to detect object in the sequence. The model selection of SVM is discussed in detail in the paper. Once the machine is trained by the samples of pre-coming frames, it can implement object recognition from the background in the following frames and thus infrared image segmentation can be performed. Practical infrared image sequence segmentation shows that the proposed method based on support vector machine needs not carry out complicated pre-processing and post-processing. Its segmentation effect is ideal with recognition rate of 99%.
出处
《光电工程》
EI
CAS
CSCD
北大核心
2005年第3期62-65,共4页
Opto-Electronic Engineering
基金
国家自然科学基金(0400067)
上海市自然科学基金(03ZR14065)
国防重点实验室基金(51476040103JW13)
关键词
支持向量机
红外图像
图像分割
目标识别
Support vector machine
Infrared image
Image segmentation
Target recognition