Coupling 2D‐wavelet decomposition and multivariate image analysis (2D WT‐MIA)

Abstract

The use of 2D discrete wavelet transform in the feature enhancement phase of multivariate image analysis is discussed and implemented in a comparative way with respect to previous publications. In the proposed approach, all the resulting subimages obtained by discrete wavelet transform decomposition are unfolded pixel-wise and midlevel data fused to a feature matrix that is used for the feature analysis phase. Congruent subimages can be obtained either by reconstruction of each decomposition block to the original pixel dimensions or by using the stationary wavelet transform decomposition scheme. The main advantage is that all possible relationships among blocks, decomposition levels, and channels are assessed in a single multivariate analysis step (feature analysis). This is particularly useful in a monitoring context where the aim is to build multivariate control charts based on images. Moreover, the approach can be versatile for contexts where several images are analyzed at a time as well as in the multispectral image analysis.Both a set of simple artificial images and a set of real images, representative of the on-line quality monitoring context, will be used to highlight the details of the methodology and show how the wavelet transform allows extracting features that are informative of how strong the texture of the image is and in which direction it varies.

A New Protocol of Computer-Assisted Image Analysis Highlights the Presence of Hemocytes in the Regenerating Cephalic Tentacles of Adult Pomacea canaliculata

“…As a consequence of these negligible shape and color differences, the levels of false positivity required a time-consuming manual elimination of false-positive identifications. New plug-ins are currently under development, by using wavelet filter to fine tuning of small differences in color-textural patterns [ 52 ], in order to automatically count also the hemocytes excluded from this analysis. Conversely, due to the specific staining and color distribution, granular Group II hemocytes were identified with high precision and accuracy ( Supplementary Table S1 ).…”

N‐way modeling for wavelet filter determination in multivariate image analysis

“…This way of performing can be included under the multi‐resolutional multivariate image analysis (MIA) framework , because we select, for some wavelet and filter length, the scales that better characterize some feature or normal operating condition image, which can also be extended for selecting those scales (and directions) that maximize some predictive ability.…”

A novel proposal to investigate the interplay between the spatial and spectral domains in near-infrared spectral imaging data by means of Image Decomposition, Encoding and Localization (IDEL)

“…Still, one possible approach is multivariate image analysis (MIA), where the unfolded imaging data is augmented with pixel-neighbor information, to incorporate local-spatial information before it is analyzed with multivariate analysis tools, such as principal component analysis (PCA) or partial least squares (PLS) regression [11][12][13]. MIA has been originally proposed for RGB images [11,13] then extended to multi-channel images [14] and only recently to spectral images [11]. However, the number of neighboring pixels increases rapidly with the distance (or window size in pixels) from the center pixel at which to consider the neighborhood, and this applies to all spectral channels, making the data unmanageable in some cases.…”

“…Some work has also been done on image segmentation, with the integration of the spectral domain [19], as well as utilizing the spectral and spatial domain, interactively switching between the two modes [20]. The analysis of textural features in spectral imaging has also been explored, by using the wavelet transform (WT) [14,[21][22][23][24][25][26]. These analyses i) use a MIA-like approach, where the local spatial information is extracted by WT, and 2D-WT sub-images are then analyzed by J o u r n a l P r e -p r o o f 4 multivariate analysis [14,[21][22][23], either on each single sub-image [21,22] or on the entire sets [14,[23][24]; ii) exploit 3D-WT on the imaging data cube [25] or iii) fuse the 2D-WT sub-images obtained at each spectral channel [26].…”