T-Test Based Adaptive Random Walk Segmentation Under Multiplicative Speckle Noise Model

Bian, Ang; Jiang, Xiaoyi

doi:10.1007/978-3-319-54427-4_41

Cited by 4 publications

(20 citation statements)

References 13 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Since we only have one sample per pixel (i.e., the actual image value) we assume that pixels in the neighborhood are from the same distribution. This assumption is also made in previous work [4,5]. subsection 3.3 is concerned with the construction of these neighborhoods.…”

Section: Weight Functionmentioning

confidence: 92%

“…There are two major works we consider as direct prior work for noise model incorporation into random walker weight functions [4,5]. In [4] additive Gaussian noise with constant global variance is assumed and the PDF of the estimated local means' difference is applied as an adaptive weight function.…”

Section: Related Workmentioning

confidence: 99%

“…In [4] additive Gaussian noise with constant global variance is assumed and the PDF of the estimated local means' difference is applied as an adaptive weight function. In [5], the signaldependent local Gaussian model with variable regional variances is assumed for the multiplicative speckle noise (with additive Gaussian noise and Loupas noise as special cases). A statistical T-test based weight function is proposed.…”

Section: Related Workmentioning

confidence: 99%

“…Grady [16] used the simple Gaussian weight function, chosen for empirical reasons, which has become common practice and has also been adopted by the recent developments of random walker segmentation [11,12,17,21,40,47]. However, Bian et al [4,5] showed that the optimal choice for its configurable parameter β highly influences the segmentation performance and depends on the image conditions. The optimal value varies even within rather homogeneous datasets and different regions of the same image.…”

Section: Introductionmentioning

confidence: 99%

“…This approach avoids the critical parameter. While previous methods on alternative weight functions work for specific noise models (additive Gaussian noise [4] and multiplicative speckle noise [5]), solutions for other noise models and in particular multi-channel images and noise are yet missing. Our work extends the application spectrum of adaptive random walker methods to many domains (e.g., biomedical imaging, remote sensing), where imaging modalities with different noise models are of high importance.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

A Bhattacharyya Coefficient-Based Framework for Noise Model-Aware Random Walker Image Segmentation

Drees¹,

Florian²,

Bian³

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

One well established method of interactive image segmentation is the random walker algorithm. Considerable research on this family of segmentation methods has been continuously conducted in recent years with numerous applications. These methods are common in using a simple Gaussian weight function which depends on a parameter that strongly influences the segmentation performance. In this work we propose a general framework of deriving weight functions based on probabilistic modeling. This framework can be concretized to cope with virtually any well-defined noise model. It eliminates the critical parameter and thus avoids time-consuming parameter search. We derive the specific weight functions for common noise types and show their superior performance on synthetic data as well as different biomedical image data (MRI images from the NYU fastMRI dataset, larvae images acquired with the FIM technique). Our framework can also be used in multiple other applications, e.g., the graph cut algorithm and its extensions.

show abstract

Section: Weight Functionmentioning

confidence: 92%