Name
Abstract | ||
|---|---|---|
This paper presents a wavelet-based method for determining a structural primitive in an image with regular texture. The overcomplete wavelet transforms and autocorrelation func- tion of its scaling image are used. An efficient peak-finding method is developed to detect peaks in the autocorrelation function for constructing two displacement vectors from which a parallelogram is formed as the structural primitive of the regular texture. Placement of the primitive in the image gives the structural component of the textured image. Image textures can be defined as repetitive patterns of vari- ous compositions including intensities, colors, and features, etc.. The extraction of repetitive patterns plays an impor- tant role in texture analysis relevant many practical applica- tions ranging from the automatic inspection of textile/carpet weaving patterns to examination of nanoscale structure reg- ularity. The 2-D autocorrelation operation is widely used for repetitive texture extraction, where peaks in the autocorrela- tion function of a textured image characterize repetitive pat- terns. Fig. 1 shows an example image containing repetitive patterns, its 2-D autocorrelation function, and a horizontal profile of the 2-D autocorrelation function passing through its center. In general, the autocorrelation function of an im- age is quite ragged and a smoothing operation is needed so that the relevant peaks can be detected effectively. Lin, et al., (4) used a Gaussian filter with an experimentally determined threshold and carried out detection of relevant peaks through a generalized Hough transform which is computation-intensive. The structural primitive of the regular texture can be extracted based on a parallelogram constructed by a pair of displace- ment vectors linking three peak locations. In this paper we present a wavelet-based approach to extract a structural prim- itive in the image texture. The biorthogonal wavelet trans- form is used to decompose an image into a number of mul- tiresolution subbands. The scaling image ( band) at a (a) Carpet image |
Year | DOI | Venue |
|---|---|---|
2001 | 10.1109/ICIP.2001.958077 | Image Processing, 2001. Proceedings. 2001 International Conference |
Keywords | Field | DocType |
correlation methods,image resolution,image texture,wavelet transforms,autocorrelation function,displacement vectors,image texture,overcomplete wavelet transforms,parallelogram,peak-finding method,regular texture,scaling image,structural primitive,wavelet analysis | Computer vision,Parallelogram,Pattern recognition,Computer science,Image texture,Low-pass filter,Artificial intelligence,Image resolution,Scaling,Autocorrelation,Wavelet transform,Wavelet | Conference |
Volume | ISBN | Citations |
3 | 0-7803-6725-1 | 4 |
PageRank | References | Authors |
0.68 | 5 | 3 |
Name | Order | Citations | PageRank |
|---|---|---|---|
| Jennting T. Hsu | 1 | 4 | 0.68 |
| Li-Chang Liu | 2 | 27 | 7.58 |
| Ching-chung Li | 3 | 383 | 65.47 |