Overview
The Texture Filter Maps Module, part of the Data Augmentation suite, generates spatial context from raster datasets by analyzing the local texture around each pixel. It uses a subset of Haralick filters to extract four key texture metrics: Contrast, Correlation, Energy, and Entropy. These are calculated from a gray-level co-occurrence matrix (GLCM), which measures how pixel values relate to their neighbours.
Rather than interpreting a single pixel in isolation, the Module evaluates patterns in the surrounding area. This local window-based analysis captures variation, repetition, and structure across the dataset. The output is a set of four raster grids, one for each filter, that describe texture properties across the input raster.
Each of the four output grids describes a different textural property:
Contrast: Shows how different the light and dark parts of the grid are. High contrast means large differences between neighbouring values.
Correlation: Measures how much a pixel’s brightness is related to its neighbors. High correlation appears in smooth, uniform zones.
Energy: Reflects how consistent or repeating the texture is. High energy appears in structured, patterned data.
Entropy: Represents randomness. High entropy is found in noisy or chaotic data regions.
This method allows DORA to better recognize patterns in geology that are defined by texture or spatial variation rather than individual values. Texture Filter Maps are especially useful when image-like properties in the input data, such as contrast or uniformity, carry geological significance.
Topic | Summary |
Module Name | Texture Filter Maps |
Purpose | Derives texture features from raster grids using GLCM‑based metrics |
Input Format | Raster |
Recommended Data | Geophysical Data; also relevant to other data types with uniform coverage and relevance to exploration, like DEM or remote sensing. |
Output Format | 4 separate Rasters: Contrast, Correlation, Energy, Entropy |
Key Parameters | AOI, input raster, window size |
Processing Summary | Applies Haralick filters over a moving window to generate 4 texture outputs |
Typical Use Cases | Highlighting texture changes in geophysics; identifying structured or chaotic zones |
Validation or QC | Not applicable (unsupervised feature engineering) |
Common Pairings | Computer Vision Maps, Feature Extraction Maps, VRIFY Prospectivity Scores (VPS) |
Notable Output Notes |
|
How It’s Used in Exploration
Texture Filter Maps are typically applied to geophysical raster data such as magnetics, gravity, or resistivity. These datasets often contain subtle textural variations that may relate to geological processes, structural features, or lithological contacts. Applying texture filters helps reveal these patterns in a consistent and quantitative way.
Within DORA, the Module is used to generate features that improve the quality of machine learning inputs. Since it creates four separate outputs from a single input, it expands the number of texture-informed variables without increasing the number of raw data layers. This is especially useful when input limits are reached or when SHAP analysis highlights that spatial context could improve model performance.
Because the filters examine each pixel’s surrounding neighbourhood rather than the pixel alone, the Module captures characteristics like:
Strong contrast zones such as edges, faults, or abrupt boundaries
Highly correlated regions such as uniform geology or parallel structures
Energetic textures with repetitive patterns or smooth surfaces
High-entropy areas where the geology appears noisy or chaotic
These spatial features often align with changes in physical properties that are difficult to detect using single-pixel approaches.
Value and Benefits
Texture Filter Maps provide valuable insight into the spatial structure of geophysical datasets. They help exploration teams identify areas where geology is consistent, transitional, or disrupted, which can be important indicators of mineralization or structural control.
The Module is efficient to use, requiring only one input raster to produce four distinct outputs. This makes it a cost-effective way to expand the input feature stack. It is also especially helpful when SHAP analysis shows that a raw geophysical input is important but was not included in the final DORA run due to input constraints. Running the Texture Filter Map on that layer can help bring its spatial context into the model without needing to include the original raster.
By focusing on the relationships between pixels, rather than just their values, Texture Filter Maps help DORA recognize more complex geological signals. This supports better predictive performance, more informed interpretations, and greater confidence in exploration targeting decisions.
Learn more:
Still have questions?
Reach out to your dedicated DORA contact or email support@VRIFY.com for more information.