Gabor kernels for affine-invariant object recognition
Chapter 13 of the book ``Gabor analysis and algorithms (\textit{H. G. Feichtinger} and \textit{T. Strohmer}, eds.), written by J. Ben-Arie and Z. Wang is concerned with the application of Gabor analysis to object recognition. The 3D-pose invariant object recognition is achieved by correlating the image with a novel configuration of modulated Gaussian kernels (multi-window Gabor scheme) and extracting local spectral signatures. The local spectral signature of each image patch is then matched against a set of iconic models using multidimensional indexing in the frequency domain. Affine-invariance of the signatures is obtained by a new configuration of modulated Gaussian kernels with modulation in two orthogonal axis. The indexing method provides robustness in partial distortion, background cutter, noise, illumination effects and lower image resolution. The authors show that their method works with high reliability in experiments over a wide range of slant, scale, swing and tilt with a dataset of 26 gray-levels and infra-red models. For the entire collection see [Zbl 0890.42004].
- Geometrically invariant color, shape and texture features for object recognition using multiple kernel learning classification approach
- Gabor wavelet networks for object representation
- Invariant object detection based on evidence accumulation and Gabor features
- scientific article; zbMATH DE number 1950300 (Why is no real title available?)
- Quadratic Gabor filters for object detection
- Object recognition using Gabor co-occurrence similarity
- scientific article; zbMATH DE number 2087648 (Why is no real title available?)
- Brief review of invariant texture analysis methods
This page was built for publication: Gabor kernels for affine-invariant object recognition
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4373889)
