Tangwei Liu

Gradient vector flow with mean shift for skin lesion segmentation

Image segmentation is an important task in the analysis of dermoscopy images since the extraction of skin lesion borders provides important cues for accurate diagnosis. In recent years, gradient vector flow based algorithms have demonstrated their merits in image segmentation. However, due to the compromise of internal and external energy forces within the partial differential equation these methods commonly lead to under- or over-segmentation problems. In this paper, we introduce a new mean shift based gradient vector flow (GVF) algorithm that drives the internal/external energies towards the correct direction. The proposed segmentation method incorporates a mean shift operation within the standard GVF cost function. Theoretical analysis proves that the proposed algorithm converges rapidly, while experimental results on a large set of diverse dermoscopy images demonstrate that the presented method accurately determines skin lesion borders in dermoscopy images.

Level set image segmentation with Bayesian analysis

Classical level set methods easily suffer from deficiency in the presence of noise and other significant edges adjacent to the real boundary. This problem has not been effectively solved in the research community. In this paper, we propose an improved energy function to tackle this problem by continuously rectifying the deviation of the level set function according to the signed distance function. This is achieved using an expectation-maximisation algorithm. Experimental work shows the proposed framework outperforms the classical level set algorithms in accuracy and efficiency of image segmentation.

Improving image segmentation by gradient vector flow and mean shift

The classical gradient vector flow (GVF) method suffers from deficiency in the presence of other significant edges adjacent to the real boundary. In this paper, we propose an improved energy function to challenge this problem by consistently reducing the Euclidean distance between the inspected centroid of the real boundary and the estimated one of the snake. Experimental work shows the proposed framework outperforms the classical GVF algorithm in different circumstances.

Skin lesion segmentation using an improved snake model

Accurate identification of lesion borders is an important task in the analysis of dermoscopy images since the extraction of skin lesion borders provides important cues for accurate diagnosis. Snakes have been used for segmenting a variety of medical imagery including dermoscopy, however, due to the compromise of internal and external energy forces they can lead to under- or over-segmentation problems. In this paper, we introduce a mean shift based gradient vector flow (GVF) snake algorithm that drives the internal/external energies towards the correct direction. The proposed segmentation method incorporates a mean shift operation within the standard GVF cost function. Experimental results on a large set of diverse dermoscopy images demonstrate that the presented method accurately determines skin lesion borders in dermoscopy images.

Segmentation of optic disc in retinal images using an improved gradient vector flow algorithm

Image segmentation plays an important role in the analysis of retinal images as the extraction of the optic disk provides important cues for accurate diagnosis of various retinopathic diseases. In recent years, gradient vector flow (GVF) based algorithms have been used successfully to successfully segment a variety of medical imagery. However, due to the compromise of internal and external energy forces within the resulting partial differential equations, these methods can lead to less accurate segmentation results in certain cases. In this paper, we propose the use of a new mean shift-based GVF segmentation algorithm that drives the internal/external energies towards the correct direction. The proposed method incorporates a mean shift operation within the standard GVF cost function to arrive at a more accurate segmentation. Experimental results on a large dataset of retinal images demonstrate that the presented method optimally detects the border of the optic disc.

Towards efficient registration of medical images

In this paper we propose a Bayesian based mutual information technique for image registration, combined with an established affine transformation model. Classical affine models allow the images to be approximately aligned. However, inefficiency and inaccuracy has appeared when using these affine models in rigorous circumstances, such as low-resolution images. To challenge this problem, we conduct mutual information measures with importance sampling to the images in an attempt to simulate the probability distribution of intensity similarity across the images. The entire registration adopts a stopping criterion as discovered in the context of differential equations. Finally, experimental results demonstrate the favorable performance of the proposed algorithm.

A hybrid framework for image segmentation

This paper presents a new approach for image segmentation by combining the classical gradient vector flow (GVF) algorithm with mean shift. Due to the dependence on the gradient vectors of an edge map, the classical GVF is sensitive to the shape irregularities, and hence the snake cannot be ideally located on the concave boundaries. We propose an improved representation of the internal energy force by reducing the Euclidean distance between the guessed centroid and the estimated one of the snake. Experimental work shows the performance of this approach in different tests.

Image restoration and detail preservation by Bayesian estimation

In this paper, we present a novel noise suppression and detail preservation algorithm. The test image is firstly pre-processed through a multiresolution analysis employing the discrete wavelet transform. Then, we apply a fast and robust total variation technique, incorporating a statistical representation in the style of maximum likelihood estimation. Finally, we compare this proposed approach to current state-of-the-art denoising methods using synthetic and real images. The results demonstrate encouraging performance of our algorithm.

Tracking non-rigid objects in video sequences

The recently proposed color based tracking systems are unable to properly adapt the ellipse that represents an object to be tracked. This most likely leads to inaccurate descriptions of the object in the later application. This paper presents a Lagrangian based method in order to discover a regularizing component for the covariance matrix. Technically, we intend to reduce the residuals between the estimated probability distribution and the expected one. We argue that, by doing this, the shape of the ellipse can be properly adapted in the tracking stage. Experimental results show that the proposed method has favorable performance in shape adaption and object localization.

RECOVERY OF NONRIGID STRUCTURES FROM 2D OBSERVATIONS

We present a new method for simultaneously determining three-dimensional (3D) motion and structure of a nonrigid object from its uncalibrated two-dimensional (2D) data with Gaussian or non-Gaussian distributions. A nonrigid motion can be treated as a combination of a rigid component and a nonrigid deformation. To reduce the high dimensionality of the deformable structure or shape, we estimate the probability distribution function (PDF) of the structure through random sampling, integrating an established probabilistic model. The fitting between the observations and the estimated 3D structure will be evaluated using the pooled variance estimator. The recovered structure is only available when the 2D feature points have been properly corresponded over two image frames. Applications of the proposed method to both synthetic and real image sequences are demonstrated with promising results.