Combination of morphometric and colorimetric has been applied for feature extraction in detection algorithm. In noisy background, this CV technique facing wrong identification results as well as ML algorithm being developed to tackle pixels cluster into salient image regions.
Dear Mohd,
Latest computer vision solutions based on dense computational approaches (as opposed to sparse ones) allow to avoid direct image feature extraction. These techniques are more relevant to robust estimation.
It would be interesting to specify whether you are referring to photometric or stereophotogrammetry techniques.
Regards
Thank u for your answer.
This scope is focused more on photometric, referred to colorimetric approach.
RGB images not involving any spectrometer, but using normal camera
Ok, I got it! My first post was specifically about multi-view imaging (photometric-based as well as stereo-based)...
From my viewpoint, it is not so much a problem of limitations as a choice of ad-hoc processing solutions.
The denoising of images (in the sense of image restoration) acquired by means of a physical imaging device is inherently more a reconstruction problem (related to inverse problem solving) than a simple filtering problem (separation of both roughness & waviness components). Image reconstruction/restoration techniques have made tremendous progress in recent years and contribute in an essential way to their subsequent analysis and interpretation.
Follow:
- Vonesch et al., " The Colored Revolution of Bioimaging ", 2006 - http://bigwww.epfl.ch/publications/vonesch0601.pdf
- Vonesch et al., "A Fast Thresholded Landweber Algorithm for Wavelet-Regularized Multidimensional Deconvolution ", 2008 - http://bigwww.epfl.ch/publications/vonesch0801.pdf
- Lefkimmiatis, et al., " Hessian-Based Norm Regularization for Image Restoration with Biomedical Applications ", 2011 -
Article Hessian-Based Norm Regularization for Image Restoration With...
- Lefkimmiatis et al., " Hessian Schatten-Norm Regularization for Linear Inverse Problems ", 2012 - http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.721.7180&rep=rep1&type=pdf
In what concerns color image segmentation, I suggest such so-called “Tree of Shapes for Multivariate Images” solution. Such a solution is invariant to contrast changes and to contrast inversion. Is is based on an hierarchical representation of the image in terms of the inclusion of its level lines.
Moreover, it is well-adapted in the absence of any assumption on the object/background layout. Follow:
- Carlinet et al., " MToS: A Tree of Shapes for Multivariate Images ", 2015 - https://hal.inria.fr/hal-01474835/document
- Carlinet et al., " A Color Tree of Shapes with Illustrations on Filtering, Simplification, and Segmentation ", 2015 - https://hal.inria.fr/hal-01476208/document
- Xu et al., " Morphological Filtering in Shape Spaces : Applications using Tree-Based Image Representations ", 2012 - https://arxiv.org/pdf/1204.4758.pdf
- Xu et al., " Connected Filtering on Tree-Based Shape-Spaces ", 2016 - https://hal.archives-ouvertes.fr/hal-01162437/document
- Xu et al., " Salient Level Lines Selection Using the Mumford-Shah Functional ", 2013 - https://hal.archives-ouvertes.fr/hal-00831032/document
Hoping it will be helpful to you.
Regards
Interesting documents. Thank u @Stephane Breton for your assistance
- Badges
- Science topic
- Similar topics
- Computer Science
- Images