期刊文献+

航空图像中阴影区域的检测 被引量:17

The detection of shadow region in aerial image
在线阅读 下载PDF
导出
摘要 航空图像中的阴影区域可以提供大量的关于目标形状、相对位置、表面特性、高度的相关信息,本文首先对阴影的特征进行了研究,认为阴影对图像的影响是使所投射地表的表面光照强度降低,但不改变原有地表,目标表面的光照特征,如纹理特征及光照强度方向统计特征等。阴影区域内的灰度方差一般小于其他非阴影区域,不同阴影区域之间的灰度值具有较强的一致性。然后,本文利用基于二维直方图的最大熵门限化方法求得边缘图像,并对边缘图像进行了细化和跟踪,删除那些对比很弱、或者虽然对比度较强,但是不满足光照条件的边缘,得到阴影边缘。在门限化过程中使用了邻域函数和过渡区域,既保留了弱边缘,又最大程度地抑制了噪声对边缘的干扰。根据阴影边缘构出假想的阴影区域,并统计这些区域的灰度直方图,求得阴影区域的灰度区间。根据该灰度区间可以得到阴影区域,这些区域具有对比度较高、至少有一侧比较规则、光学性质类似的特点。 The shadow regions in aerial image provide lots of information about the appearance, relative position and surface feature of object, and they are the obvious evidences of objects higher than the surrounding background. In this paper the features of shadow are firstly studied. Shadow in an image reduces the surface illumination intensity of the cast area, but not the illumination features of the surface of the earth and the object. For example, the' texture feature and the directional statistical features of illumination intensity are almost not changed. The gray value covariance of the shadow region is usually less than other regions. The gray values of points in different shadow region are rather consistent with each other. Then the edge image is obtained through the maximum entropy thresholding of 2D histogram. The maximum difference of gray value of neighboring window and the transition region are employed to retain the weak edges and wipe off the noise's influence to maximum degree. Then the edge image is thinned and independent edge points are linked to small line fragments. Then fragments that have weaker contrast and that have stronger contrast but the direction is not consistent with the illumination are removed. The remaining edges are named, shadow edges. The supposing shadow regions can be constructed along the shadow edges. After calculating the histogram of gray value, the gray distribution of shadow region can be acquired, and then the shadow region. The shadow region achieved in this paper has the features that the contrasts are relatively high, there is at least one side that is considerable regular, and that the optics qualities of shadow regions are similar.
出处 《信号处理》 CSCD 2002年第3期228-232,共5页 Journal of Signal Processing
关键词 航空图像 阴影区域 检测 二维直方图 最大熵门限化 最大领域差 细化 2D histogram Maximum entropy thresholding Line detection Maximum gray difference of neighbor window Thinning
  • 相关文献

参考文献8

  • 1[1]Barnes N. And Liu Z.Q., knowledge-based shape from shading. Int. of. of Pattern recognition and Artificial Intelligence, 1999, 13(1)1~23.
  • 2[2]Ortega A. and shah M., From shape from shading to object recognition, Int. of. of P-attern Recognition and Artificial Intelligence, 1998, 12(2):191~208.
  • 3[3]J.B. Thirion. Realistic 3D simulatron of shapes and shadows for image processing, CVGIP, 1992,54(1)82~90.
  • 4[4]C.Wang, L. Huang,and A.Rosenfeld. Detecting clouds and cloud shadows on aerial photographd, PRL, 1991, 12(1): 55-64.
  • 5[5]A. Huertas and R. Nevatia, Detecting Buildings in Aerial Images, CVGIP, 1988, 41 (1): 131~152.
  • 6[6]Feng L. Suen C. Y., Tang Y.Y. and Yang L.H. Edge extraction of images by reconstruction using wavelet decomposition details at different resolution levels. Int. J.of Pattern Recognition and Artificial Intelligence,2000,14(6): 779~793.
  • 7[7]Cheng H. D., Y.H. Chen, and X.H. Jiang. Thresholding using two-dimensional histogram and fuzzy entropy principle. T-IP,2000,9(4):732~725.
  • 8[8]Fu Chang, Ya-Ching Lu, and T. Pavlidis. Feature Analysis using line sweep thinning algorethm, T-PAMI, 1999,21(2): 145~158.

同被引文献135

引证文献17

二级引证文献162

相关作者

内容加载中请稍等...

相关机构

内容加载中请稍等...

相关主题

内容加载中请稍等...

浏览历史

内容加载中请稍等...
;
使用帮助 返回顶部