AI-ordlista
Den kompletta ordlistan över AI
Texture Segmentation
Process of partitioning an image into homogeneous regions based on textural features, without considering color or intensity.
Co-occurrence Matrix (GLCM)
Statistical method for texture analysis that evaluates the spatial frequency of pairs of gray levels in a given direction and distance.
Texture Map
Two-dimensional representation where each pixel encodes a local texture measure, serving as the basis for texture segmentation.
Local Entropy
Measure of the unpredictability of gray levels within a neighborhood window, used to quantify the textural complexity of a region.
Multi-resolution Segmentation
Hierarchical approach analyzing textures at different spatial scales to improve the robustness and accuracy of region delineation.
LBP Texture Descriptors
Local Binary Pattern, an operator that encodes micro-texture by comparing each pixel to its neighbors, creating a robust histogram for segmentation.
Markov Models
Probabilistic approach modeling the spatial dependency of pixels to characterize and segment textures based on random fields.
Textural Homogeneity
Segmentation criterion measuring the similarity of textural properties within a region, often based on the variance of descriptors.
Anisotropic Filtering
Smoothing technique that preserves edges while reducing intra-regional noise, essential for preparing images for texture segmentation.
Markov Random Fields (MRF)
Probabilistic model where the state of a pixel depends on its neighbors, used to impose a spatial coherence constraint in texture segmentation.
2D Spectrogram
Spatial frequency representation of an image obtained by local Fourier transform, revealing the periodic patterns characteristic of textures.
Texture Clustering Methods
Unsupervised algorithms (like K-means on texture descriptors) grouping pixels into regions sharing similar textural characteristics.
Texture Region Growing
Iterative algorithm that expands an initial region by adding neighboring pixels whose textural properties are similar to those of the seed.
Haralick Descriptors
Set of 14 statistical features (contrast, correlation, energy, homogeneity) extracted from the GLCM to finely describe texture.
Data Fusion Segmentation
Approach combining multiple texture descriptors (statistical, frequency-based, structural) for more robust and accurate segmentation.