摘要
从近红外星空图像的直方图出发,认真分析了直方图的特性,并通过对直方图低频区域图像进行高斯拟合确定了图像分割阈值。对不同阈值情况下取得的结果进行了对比。在选取8邻域内像素面积最大的10个目标作为导航目标时,该方法用的数据较少,大幅减小了计算机的工作量,从而提高了导航效率,并增快了识别速度。
In the view of a histogram of near infrared starry sky images,the characteristics of the histogram are analyzed.The threshold for image segmentation is determined by fitting the low area of the histogram with a Gaussian function.The results obtained with different thresholds are compared. When ten objects which have the largest area in the eight-neighborhood are selected as the navigation objects,less data are used.Because the calculation amount of a computer is reduced greatly,both navigation effectiveness and recognition speed are improved.
出处
《红外》
CAS
2013年第5期42-45,共4页
Infrared
关键词
高斯拟合
阈值分割
邻域面积
gaussian fitting
threshold segmentation
neighborhood area
