Name
Abstract | ||
|---|---|---|
By incorporating the local statistics of an image, a semi-causal non-stationary autoregressive random field can be applied to a non-stationary image for segmentation. Because this non-stationary random field can provide a better description of the image texture than the stationary one, an image can be better segmented. Besides low-order dependence among pixels in image for above-mentioned texture random field, the paper also introduces high-order dependence as a new classification feature to recognize the real object. Entropy rate that depicts the high-order dependence feature can also be estimated by using random field model. The proposed technique is applied to extract urban areas from a Landsat image. |
Year | DOI | Venue |
|---|---|---|
2004 | 10.1016/j.patrec.2003.09.006 | Pattern Recognition Letters |
Keywords | Field | DocType |
texture segmentation,non-stationary random field,double random field,low-order dependence,non-stationary image,high-order dependence,random field model,high-order dependence feature,high-order feature,image texture,stochastic models,random field,non-stationary autoregressive random field,landsat image,image segmentation,stochastic model,entropy rate | Autoregressive model,Computer vision,Random field,Scale-space segmentation,Pattern recognition,Feature detection (computer vision),Image texture,Segmentation-based object categorization,Image segmentation,Pixel,Artificial intelligence,Mathematics | Journal |
Volume | Issue | ISSN |
25 | 1 | Pattern Recognition Letters |
Citations | PageRank | References |
8 | 0.53 | 17 |
Authors | ||
2 | ||
Name | Order | Citations | PageRank |
|---|---|---|---|
| Feng Li | 1 | 10 | 0.91 |
| Peng Jiaxiong | 2 | 45 | 8.03 |