GPU-accelerated image segmentation based on level sets and multiple texture features

Abstract

In this paper, we present a fast multi-stage image segmentation method that incorporates texture analysis into a level set-based active contour framework. This approach allows integrating multiple feature extraction methods and is not tied to any specific texture descriptors. Prior knowledge of the image patterns is also not required. The method starts with an initial feature extraction and selection, then performs a fast level set-based evolution process and ends with a final refinement stage that integrates a region-based model. The presented implementation employs a set of features based on Grey Level Co-occurrence Matrices, Gabor filters and structure tensors. The high performance of feature extraction and contour evolution stages is achieved with GPU acceleration. The method is validated on synthetic and natural images and confronted with results of the most similar among the accessible algorithms.

GPU-accelerated lung CT segmentation based on level sets and texture analysis

“…The prepared implementation can independently apply two level set-based deformable models: an intensity-based 3D active surface 32 and a texture-based 2D model 35 that utilises GLCM and Gabor features. Although the method is focused on 3D segmentation, the 2D model is also employed to emphasise the advantages of the fully three-dimensional approach.…”

“…The intensity-based version 32 defines the data term as: where I ( p ) is the image intensity in p , while T is the intensity target (the mean intensity in the initial regions) and is the tolerance. In the 2D case, the proposed method can also employ a multi-feature data term that takes into consideration the features (denoted as a set M ) generated for the given volume (please refer to 35 for the details on feature generation and selection). For each point p , a subset of features is defined as: where and are the feature’s mean and standard deviation inside the initial regions, m ( p ) is the value of feature m in the point p and is a user-defined sensitivity parameter.…”

Performance evaluation of spatial fuzzy C-means clustering algorithm on GPU for image segmentation

“…Many image segmentation techniques try to discharge their computations to GPU hardware. Researchers in [ 26 , 38 ] performed a level set medical image segmentation on GPU using efficiently CUDA programming environment. They have lowered significantly the execution time from about 8 s to less than 0.5 s for the level set method.…”

Intelligent Control Techniques for the Detection of Biomedical Ear Infections

“…It is quite difficult to do quantitative analysis on medical photos because of the rich texture and fuzzy border present in these images. Because of this, it would be required to conduct a performance review of these procedures [ 8 , 15 , 16 ]. This effort resulted in the development of many segmentation algorithms, which are explained further down in this section.…”

“…ree different wavelengths were used: 450 nm, 530 nm, and 630 nm. ree distinct wavelengths of light were used to capture the spectroscopic images: 450 nanometers, 530 nanometers, and 630 nanometers [4][5][6][7][8]. It has been suggested that using such pictures in the evaluation of the tympanic membrane (TM) may make it easier to determine the boundaries of the TM and segment it with high precision [4][5][6][7][8].…”