人工智能
计算机视觉
图像分割
计算机科学
图像纹理
亮度
相似性(几何)
图像处理
分割
特征检测(计算机视觉)
二值图像
尺度空间分割
序列(生物学)
区域增长
模式识别(心理学)
图像(数学)
光学
物理
生物
遗传学
作者
Yilun Cheng,Fengfu Tan,Feng He,Laian Qin,Zhigang Huang,Zaihong Hou
摘要
When the camera imaging method and the detector array method are used to obtain the laser spot image after atmospheric transmission, the image sequence will contain the actual spot image information and the background information of the test system. During the data processing, the image sequence needs to be segmented to eliminate Image background information firstly. Considering that the brightness, contrast, and structure of the images collected by the system will change significantly before and after the laser irradiation measurement system, a method of image structure similarity is proposed to segment the background information of the image sequence. The method of dynamic template is used to calculate the similarity of the light spot sequence, and the similarity threshold is set to complete the segmentation and elimination of the background information of the image. Experiments show that the background segmentation method of image sequence based on structural similarity can quickly and accurately achieve segmentation of background information. The research results can provide a certain theoretical basis and method accumulation for establishing a perfect data processing method in the laser parameter measurement system.
科研通智能强力驱动
Strongly Powered by AbleSci AI