Separation of image parts using 2-D parallel form recursive filters

被引:0
|
作者
Sivaramakrishna, R
机构
[1] Department of Electrical and Computer Engineering, University of Manitoba, Winnipeg
关键词
D O I
10.1109/83.481686
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
This correspondence deals with a new technique to separate objects or image parts in a composite image. A parallel form extension of a 2-D Steiglitz-McBride method is applied to the discrete cosine transform (DCT) of the image containing the objects that are to be separated. The obtained parallel form is the sum of several filters or systems, where the impulse response of each filter corresponds to the DCT of one object in the original image, Preliminary results on an image with two objects show that the algorithm works well, even in the case where one object occludes another as well as in the case of moderate noise.
引用
收藏
页码:175 / 178
页数:4
相关论文
共 50 条