Abstract:The analysis of the spatial arrangement of colors and roughness/smoothness of figures is relevant due to its wide range of applications. This paper proposes a texture classification method that extracts data from images using the Hilbert curve. Three information theory quantifiers are then computed: permutation entropy, permutation complexity, and Fisher information measure. The proposal exhibits some important properties: (i) it allows to discriminate figures according to varying degrees of correlations (as measured by the Hurst exponent), (ii) it is invariant to rotation and symmetry transformations, (iii) it can be used either in black and white or color images. Validations have been made not only using synthetic images but also using the well-known Brodatz image database.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is to develop a robust and efficient texture classification method to overcome the problems encountered by traditional methods in processing image textures, such as high computational complexity and sensitivity to image quality (such as noise, blurring and color changes). Specifically, the paper proposes an information quantization method based on Hilbert curve paths for texture discrimination. The main objectives of this method include: 1. **Distinguishing images with different degrees of correlation**: By measuring the Hurst exponent, images can be classified according to their degree of correlation. 2. **Invariance to rotation and symmetry transformation**: Ensure that the algorithm can still classify accurately after the image is rotated or mirror - symmetrically transformed. 3. **Applicable to black - and - white and color images**: This method can be effectively applied to both black - and - white and color images. ### Method Overview To achieve the above objectives, the paper proposes a two - step method: 1. **Conversion of images to one - dimensional time series**: Use the Hilbert curve to convert a two - dimensional image into a one - dimensional time series. The Hilbert curve is a space - filling curve that can traverse each pixel in the image once while maintaining spatial locality. 2. **Calculation of information - theoretic quantifiers**: Extract three information - theoretic quantifiers from the converted time series: - **Permutation Entropy (PE)** - **Permutation Statistical Complexity (PSC)** - **Fisher Information Measure (FIM)** These quantifiers can capture the intrinsic characteristics of the image and are used for classifying and distinguishing different texture types. ### Verification and Advantages The paper verifies the effectiveness of this method in the following ways: - Test using synthetic images generated by self - similar multifractal surfaces and Brownian surfaces. - Use the well - known Brodatz image database for verification, which is widely used in texture classification research. The experimental results show that this method can not only accurately classify different types of textures, but also shows good robustness in cases such as image rotation, mirror - symmetric transformation and color change. ### Main Contributions The main contributions of this method are: 1. **Elimination of directional bias**: Compared with traditional row/column scanning, the Hilbert curve can represent the texture features of the image more evenly. 2. **Efficient space - filling**: Utilize the non - privileged - direction property of the Hilbert curve to ensure that each pixel is visited only once, avoiding data redundancy. 3. **Powerful information - theoretic quantifiers**: By combining permutation entropy, permutation statistical complexity and Fisher information measure, a robust and information - rich texture classification framework is provided. 4. **Robustness to rigid - body transformations**: This method is stable under rigid - body transformations such as image rotation and mirror - symmetry. 5. **Scalability**: This method can be naturally extended to higher dimensions for classifying n - dimensional patterns. In summary, this paper proposes a novel and effective texture classification method, aiming to improve the accuracy and robustness of texture analysis and has broad application prospects.

Texture Discrimination via Hilbert Curve Path Based Information Quantifiers

Image Retrieval Using VQ Index Histograms in Classified DCT Domain.

Classifying Image Texture with Statistical Landscape Features

A novel feature descriptor for image retrieval by combining modified color histogram and diagonally symmetric co-occurrence texture pattern

Color Image Retrieval Based on Color-texture-edge Feature Histograms.

Local Quantization Code Histogram for Texture Classification

Statistical Landscape Features for Texture Classification

A Segmentation Method of Color Texture Image

Compact Interchannel Sampling Difference Descriptor for Color Texture Classification

Hilbert Spectrum Analysis of Piecewise Stationary Signals and Its Application to Texture Classification.

Texture Feature Extraction of Image Based on 2D Hilbert-Huang Transform and Multifractal Analysis

A non-extensive entropy feature and its application to texture classification

Texture classification using joint statistical representation in space-frequency domain with local quantized patterns

Multiresolution gray-scale and rotation invariant texture classification with local binary patterns

3D Mixed Invariant and Its Application on Object Classification

A robust texture descriptor using multifractal analysis with Gabor filter.

Texture Classification Using Local Pattern Based on Vector Quantization

A Projective Invariant for Textures

Image cognition using contour curvature statistics

A Texture Descriptor for Image Retrieval and Browsing