Christine M. Zwart

Segment Adaptive Gradient Angle Interpolation

We introduce a new edge-directed interpolator based on locally defined, straight line approximations of image isophotes. Spatial derivatives of image intensity are used to describe the principal behavior of pixel-intersecting isophotes in terms of their slopes. The slopes are determined by inverting a tridiagonal matrix and are forced to vary linearly from pixel-to-pixel within segments. Image resizing is performed by interpolating along the approximated isophotes. The proposed method can accommodate arbitrary scaling factors, provides state-of-the-art results in terms of PSNR as well as other quantitative visual quality metrics, and has the advantage of reduced computational complexity that is directly proportional to the number of pixels.

An In Vitro Study of Pulsatile Fluid Dynamics in Intracranial Aneurysm Models Treated with Embolic Coils and Flow Diverters

Although coil embolization is one of the most effective treatments for intracranial aneurysms (ICAs), the procedure is often unsuccessful. For example, an ICA may persist after coil embolization if deployed coils fail to block the flow of blood into the aneurysm. Unfortunately, the specific flow changes that are effected by embolic coiling (and other endovascular therapies) are poorly understood, which creates a barrier to the design and execution of optimal treatments in the clinic. We present an in vitro pulsatile flow study of treated basilar tip aneurysm models that elucidates relationships between controllable treatment parameters and clinically important post-treatment fluid dynamics. We also compare fluid dynamic performance across embolic coils and more recently proposed devices (e.g., the Pipeline Embolization Device) that focus on treating ICAs by diverting rather than blocking blood flow. In agreement with previous steady flow studies, coil embolization-reduced velocity magnitude at the aneurysmal neck by greater percentages for a narrow-neck aneurysm, and reduced flow into aneurysms by greater percentages at lower parent vessel flow rates. However, flow diversion reduced flow into a wide-neck aneurysm more so than coil embolization, regardless of flow conditions. Finally, results also showed that for the endovascular devices we examined, treatment effects were generally less dramatic under physiologic pulsatile flow conditions as compared to steady flow conditions. The fluid dynamic performance data presented in this study represent the first direct in vitro comparison of coils and flow diverters in aneurysm models, and provide a novel, quantitative basis to aid in designing endovascular treatments toward specific fluid dynamic outcomes.

Soft adaptive gradient angle interpolation of grayscale images

We introduce a new edge-directed interpolator based on locally defined, straight line approximations to image isophotes. The first spatial derivatives of image intensity are used to describe the behavior of pixel-intersecting isophotes in terms of their slopes. Slopes are determined by inverting a tridiagonal matrix and are forced to vary linearly from pixel to pixel. Image resizing is performed using standard, 1D interpolators along the approximated isophotes. The proposed method can accommodate arbitrary scaling factors, provides state-of-the-art results in terms of PSNR as well as other quantitative quality metrics, and has the advantage of computational complexity that is directly proportional to the number of pixels.

Biaxial Control Grid Interpolation: Reducing isophote preservation to optical flow

The importance of edge fidelity in image resizing is well established. Isophotes, or connected pixels of equal intensity, are essential to human perception of static images and interpolation methods that disrupt isophote curvature produce distracting artifacts. We introduce a new image resizing algorithm based on the principles of optical flow. The optical flow equation assumes that for every pixel in a given video frame, there exists an isointense pixel in adjacent frames. For video, this assumption implies that subsequent frames are reconfigurations of the same pixels. Here, we apply the optical flow equation to adjacent rows and columns of single images. The physical basis for optical flow in video (objects are moving) does not apply to static images. However, the the use of the optical flow equation amounts to asserting that each pixel is a member of an isophote with curvature that can be approximated locally with a straight line. Our implementation is fully separable and outperforms both traditional and recently proposed interpolators including NEDI and iNEDI.

Improved motion estimation for restoring turbulence-distorted video

Artificial displacement (the apparent motion of stationary objects) is one important component of atmospheric turbulence distortion, which has led many researchers to propose motion compensation as a solution. Defining a sufficiently dense set of motion estimates for successful restoration is challenging, particularly for time sensitive applications. We introduce a new, control grid implementation of optical flow that allows for rapid and analytical solutions to the motion estimation problem. Our results demonstrate the effectiveness of using the resulting motion field for removing articial displacements in turbulence distorted videos.

Video deinterlacing with control grid interpolation

Video deinterlacing is a key technique in digital video processing, particularly with the widespread usage of LCD and plasma TVs. This paper proposes a novel spatio-temporal video deinterlacing technique that adaptively chooses between results from segment adaptive gradient angle interpolation (SAGA), vertical temporal filter (VTF) and temporal line averaging (LA). The proposed method performs better than several popular benchmarking methods in terms of both visual quality and PSNR and requires minimal computational overhead. The algorithm performs better than existing approaches on fine moving edges and semi-static regions of videos, which are recognized as particularly challenging deinterlacing cases.

One-dimensional control grid interpolation-based demosaicing and color image interpolation

We recently reported good results with our image interpolation algorithm, One-Dimensional Control Grid Interpolation (1DCGI), in the context of grayscale images. 1DCGI has high quantitative accuracy, flexibility with respect to scaling factor, and low computational cost relative to similarly performing methods. Here we look to extend our method to the demosaicing of Bayer-Patterned images. 1DCGI-based demosaicing performs quantitatively better than the gradient-corrected linear interpolation method of Malvar. We also demonstrate effective interpolation of full color images.

Decomposed multidimensional control grid interpolation for common consumer electronic image processing applications

Abstract. Interpolation is an essential and broadly employed function of signal processing. Accordingly, considerable development has focused on advancing interpolation algorithms toward optimal accuracy. Such development has motivated a clear shift in the state-of-the art from classical interpolation to more intelligent and resourceful approaches, registration-based interpolation for example. As a natural result, many of the most accurate current algorithms are highly complex, specific, and computationally demanding. However, the diverse hardware destinations for interpolation algorithms present unique constraints that often preclude use of the most accurate available options. For example, while computationally demanding interpolators may be suitable for highly equipped image processing platforms (e.g., computer workstations and clusters), only more efficient interpolators may be practical for less well equipped platforms (e.g., smartphones and tablet computers). The latter examples of consumer electronics present a design tradeoff in this regard: high accuracy interpolation benefits the consumer experience but computing capabilities are limited. It follows that interpolators with favorable combinations of accuracy and efficiency are of great practical value to the consumer electronics industry. We address multidimensional interpolation-based image processing problems that are common to consumer electronic devices through a decomposition approach. The multidimensional problems are first broken down into multiple, independent, one-dimensional (1-D) interpolation steps that are then executed with a newly modified registration-based one-dimensional control grid interpolator. The proposed approach, decomposed multidimensional control grid interpolation (DMCGI), combines the accuracy of registration-based interpolation with the simplicity, flexibility, and computational efficiency of a 1-D interpolation framework. Results demonstrate that DMCGI provides improved interpolation accuracy (and other benefits) in image resizing, color sample demosaicing, and video deinterlacing applications, at a computational cost that is manageable or reduced in comparison to popular alternatives.

Spatiotemporal video deinterlacing using control grid interpolation

Abstract. With the advent of progressive format display and broadcast technologies, video deinterlacing has become an important video-processing technique. Numerous approaches exist in the literature to accomplish deinterlacing. While most earlier methods were simple linear filtering-based approaches, the emergence of faster computing technologies and even dedicated video-processing hardware in display units has allowed higher quality but also more computationally intense deinterlacing algorithms to become practical. Most modern approaches analyze motion and content in video to select different deinterlacing methods for various spatiotemporal regions. We introduce a family of deinterlacers that employs spectral residue to choose between and weight control grid interpolation based spatial and temporal deinterlacing methods. The proposed approaches perform better than the prior state-of-the-art based on peak signal-to-noise ratio, other visual quality metrics, and simple perception-based subjective evaluations conducted by human viewers. We further study the advantages of using soft and hard decision thresholds on the visual performance.

Control Grid Motion Estimation for Efficient Application of Optical Flow

Abstract Motion estimation is a long-standing cornerstone of image and video processing. Most notably, motion estimation serves as the foundation for many of todays ubiquitous video coding standards including H.264. Motion estimators also play key roles in countless other applications that serve the consumer, industrial, biomedical, and military sectors. Of the many available motion estimation techniques, optical flow is widely regarded as most flexible. The flexibility offered by optical flow is particularly useful for complex registration and interpolation problems, but comes at a considerable computational expense. As the volume and dimensionality of data that motion estimators are applied to continue to grow, that expense becomes more and more costly. Control grid motion estimators based on optical flow can accomplish motion estimation with flexibility similar to pure optical flow, but at a fraction of the computational expense. Control grid methods also offer the added benefit of representing motion...