3D Solid Texture Classification Using Locally-Oriented Wavelet Transforms

Abstract

Abstract-Many image acquisition techniques used in biomedical imaging, material analysis, and structural geology are capable of acquiring 3D solid images. Computational analysis of these images is complex but necessary, since it is difficult for humans to visualize and quantify their detailed 3D content. One of the most common methods to analyze 3D data is to characterize the volumetric texture patterns. Texture analysis generally consists of encoding the local organization of image scales and directions, which can be extremely diverse in 3D. Current state-of-the-art techniques face many challenges when working with 3D solid texture, where most approaches are not able to consistently characterize both scale and directional information. 3D Riesz-wavelets can deal with both properties. One key property of Riesz filterbanks is steerability, which can be used to locally align the filters and compare textures with arbitrary (local) orientations. This paper proposes and compares three novel local alignment criteria for higher-order 3D Rieszwavelet transforms. The estimations of local texture orientations are based on higher-order extensions of regularized structure tensors. An experimental evaluation of the proposed methods for the classification of synthetic 3D solid textures with alterations (such as rotations and noise) demonstrated the importance of local directional information for robust and accurate solid texture recognition. These alignment methods achieved an accuracy of 0.95 in the rotated data, three times more than the unaligned Riesz descriptor that achieved 0.32. The accuracy obtained is better than all other techniques that are published and tested on the same database.Index Terms-3D solid texture analysis, 3D texture classification, Riesz-wavelets steerability, aligned higher-order Riesz-wavelet transform, rotation-covariance.

From Local to Global: A Holistic Lung Graph Model

“…Then, the lung masks were geometrically divided into a 36-region atlas [3,5] derived from the 3D model of the human lung presented by Zrimec et al [17]. For each region r of this atlas two texturebased feature descriptors were extracted: the Fourier histograms of oriented gradients (FHOG) [13] and the locally-oriented 3D Riesz-wavelet transform (3DRiesz) [9]. These descriptors have been successful for multiple biomedical texture analysis applications [1,2].…”

“…FHOG was computed using 28 3D directions for the histogram, obtaining a 28-dimensional feature vector per image voxel v (f H (v) ∈ R 28 ). For 3DRiesz we used the 3rd-order Riesz-wavelet transform, with 4 scales and 1st-order alignment (see [9]). The feature vector for a single voxel was defined as the weighted sum of the absolute Riesz response along the 4 scales, obtaining a 10-dimensional feature vector (f R (v) ∈ R 10 ).…”

Investigating local orientation methods to segment microstructure with 3D solid texture

“…Another type of approach based on wavelets has become popular to capture and represent surface‐like singularities in multidimensional data. Recent works have proposed to use 3D steerable Riesz wavelet [14, 15] to classify solid texture patterns [16] or act as a directional filter bank [17]. The proposed wavelet frame relies on the combination of a 3D isotropic wavelet transform with the 3D Riesz operator which brings steerability to the pyramid, i.e.…”

An automatically generated texture-based atlas of the lungs

“…As texture descriptors we used the locally-oriented 3D Riesz-wavelet transform introduced by Dicente et al 15 The configuration chosen for this work obtained one of the best results for texture classification on the RFAI database: the 3rd order Riesz transform, with 4 scales and the local-orientation method based on the 1st order Riesz filters (see 15 ). This configuration provides 40-dimensional feature vectors for each voxel of the volume.…”

“…As texture features we used the previously validated locallyoriented 3D Riesz-wavelet transform. 15 Finally, we created the different regions of the atlas using a 3D extension of a superpixel 16 algorithm (supervoxels) based on spatial and texture information. In addition to this atlas, we make a synthetic solid 3D texture database available to the scientific community for training the segmentation algorithms.…”