基于区域生长的前视红外图像分割方法

红外成像制导技术以其优越的性能成为当今精确制导技术发展的主流。红外图像的精确分割是实现目标识别的基础。针对地面目标前视红外图像的特点和成像制导技术中图像分割的目的，提出了一种基于区域生长的前视红外图像分割算法，它首先在全局阈值分割的基础上选择出种子点所在区域，并在区域中定义局部灰度信息统计准则和策略选取出种子点；然后以目标模板面积作为参考，通过对分割效果的判断自动调整生长阈值进行区域生长以得到分割图像。实验结果表明，本方法分割出的目标完整准确，分割结果对基于边缘特征的目标匹配识别非常有利。

Forward-looking infrared image segmentation based on region growing

Infrared imaging guidance technology has been the main trend in the development of precision guidance technology today by reason of its predominant capability.Accurate infrared image segmentation is the basis for realizing target recognition.Aiming at the features of forward-looking infrared image and the purpose of image segmentation in imaging guidance technology,a novel segmentation algorithm based on region growing is proposed.Firstly the seed region is selected based on global threshold segmentation and partial gray scale statistics criterion and strategy is defined to choose the seeds in the region.Then regarding the area of target template as the reference,the growing threshold is adjusted automatically for region growing through the estimation of segmentation effect.Finally,the segmented image is obtained.Experimental results show that the target information gained by the proposed algorithm is complete and exact and the result greatly facilitates the target matching recognition based on edge feature.