Abstract:Feature extraction is a key step in image processing for pattern recognition and machine learning processes. Its purpose lies in reducing the dimensionality of the input data through the computing of features which accurately describe the original information. In this article, a new feature extraction method based on Discrete Modal Decomposition (DMD) is introduced, to extend the group of space and frequency based features. These new features are called modal features. Initially aiming to decompose a signal into a modal basis built from a vibration mechanics problem, the DMD projection is applied to images in order to extract modal features with two approaches. The first one, called full scale DMD, consists in exploiting directly the decomposition resulting coordinates as features. The second one, called filtering DMD, consists in using the DMD modes as filters to obtain features through a local transformation process. Experiments are performed on image texture classification tasks including several widely used data bases, compared to several classic feature extraction methods. We show that the DMD approach achieves good classification performances, comparable to the state of the art techniques, with a lower extraction time.

What problem does this paper attempt to address?

The problem this paper attempts to address is the feature extraction issue in image texture classification. Specifically, the authors propose a new feature extraction method based on Discrete Modal Decomposition (DMD) to extend existing spatial and frequency features. These new features are referred to as modal features. The main goal of the paper is to achieve comparable or even better results than existing techniques in terms of reducing computation time and improving classification performance through this new feature extraction method. ### Background of the Paper In image processing, feature extraction is a critical step aimed at reducing the dimensionality of input data by computing features that can accurately describe the original information. Existing feature extraction methods mainly include geometric methods, statistical methods, filtering or transformation methods, and texture space modeling methods. However, these methods may not efficiently extract features in certain cases, especially when dealing with complex texture images. ### Research Contributions 1. **Proposing a New Feature Extraction Method**: The authors propose a feature extraction method based on DMD, which can be used to extract modal features of images. 2. **Two DMD Feature Extraction Methods**: - **Full-scale DMD**: Directly uses the coordinates obtained from DMD decomposition as features. - **Filtered DMD**: Uses DMD modes as filters to obtain features through a local transformation process. 3. **Experimental Validation**: Experiments were conducted on multiple widely-used image texture classification datasets and compared with classical feature extraction methods (such as Haralick features, LBP features, HOG features, and DCT features). ### Experimental Results - **Classification Performance**: The DMD method achieved classification performance comparable to or even better than existing techniques on multiple datasets. - **Extraction Time**: The feature extraction time of the DMD method was superior to the other three methods across all datasets, particularly in the full-scale DMD method, where the computational burden was significantly reduced. ### Conclusion By introducing a DMD-based feature extraction method, the paper demonstrates the effectiveness and efficiency of this method in image texture classification tasks. This provides new ideas and tools for research in the fields of image processing and pattern recognition.

Modal features for image texture classification

Nonlinear Discrete Cross-Modal Hashing for Visual-Textual Data

Feature Based On Modulus Maxima Of Wavelet Frame Representation For Texture Retrieval

An improved texture image classification method based on bidimensional empirical mode decomposition

Large Margin Multi-modal Multi-task Feature Extraction for Image Classification

Bi-dimensional Variational Mode Decomposition for Surface Texture Analysis

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

Face Feature Extraction and Recognition Via Local Binary Pattern and Two-Dimensional Locality Preserving Projection

Decorrelated Local Binary Patterns for Efficient Texture Classification

Feature Extraction Using Dominant Local Texture-Color Patterns (DLTCP) and Classification of Color Images

Hyperspectral Image Classification Based on Structured Sparse Logistic Regression and Three-Dimensional Wavelet Texture Features.

Nonnegative Multiresolution Representation-Based Texture Image Classification

Simultaneous and Orthogonal Decomposition of Data Using Multimodal Discriminant Analysis

Fractal measures of image local features: an application to texture recognition

Efficient texture retrieval using multiscale local extrema descriptors and covariance embedding

Robust directional median pattern for noisy texture classification

Statistical multiscale blob features for classifying and retrieving image texture from large-scale databases

Describing Textures in the Wild

Feature discovering for image classification via wavelet-like pattern decomposition.

Median Robust Extended Local Binary Pattern for Texture Classification

Efficient Feature Extraction For Image Classification