Gerard De Haan

Sub-pixel motion estimation with 3-D recursive search block-matching

Abstract Recently the 3-D Recursive Search Block-Matching algorithm was introduced as a high quality, low-cost, true-motion estimation method suitable for critical field rate conversion applications. In thisarticle an extension of the algorithm is presented that provides a sub-pixel accuracy of the estimated motion vectors. This significantly broadens the applicability of the algorithm in the area of interlaced-to-sequential scan conversion and coding. The extension is such that it hardly adds any calculational complexity, which implies that the attractiveness of the algorithm for a VLSI implementation remains high. Even more, a simplified version of the algorithm, the Y-prediction block-matcher, is suggested that offers sub-pixel accuracy, a large range of motion vectors, and an extremely low complexity requiring only four candidate vectors per block. An evaluation of this estimator is included in the paper.

Subpixel image scaling for color‐matrix displays

The perceived resolution of matrix displays increases when the relative position of the color subpixels is taken into account. Subpixel-rendering algorithms are being used to convert an input image to subpixel-corrected display images. This paper deals with the consequences of the subpixel structure and the theoretical background of the resolution gain. We will show that this theory allows a low-cost implementation in an image scaler. This leads to high flexibility, allowing different subpixel arrangements and a simple control over the trade-off between perceived resolution and color errors.

Motion Robust Remote-PPG in Infrared

Current state-of-the-art remote photoplethysmography (rPPG) algorithms are capable of extracting a clean pulse signal in ambient light conditions using a regular color camera, even when subjects move significantly. In this study, we investigate the feasibility of rPPG in the (near)-infrared spectrum, which broadens the scope of applications for rPPG. Two camera setups are investigated: one setup consisting of three monochrome cameras with different optical filters, and one setup consisting of a single RGB camera with a visible light blocking filter. Simulation results predict the monochrome setup to be more motion robust, but this simulation neglects parallax. To verify this, a challenging benchmark dataset consisting of 30 videos is created with various motion scenarios and skin tones. Experiments show that both camera setups are capable of accurate pulse extraction in all motion scenarios, with an average SNR of +6.45 and +7.26 dB, respectively. The single camera setup proves to be superior in scenarios involving scaling, likely due to parallax of the multicamera setup. To further improve motion robustness of the RGB camera, dedicated LED illumination with two distinct wavelengths is proposed and verified. This paper demonstrates that accurate rPPG measurements in infrared are feasible, even with severe subject motion.

13.4: Subpixel Image Scaling for Color Matrix Displays

The apparent resolution of matrix displays is increased when the relative position of the color subpixels is taken into account. This paper shows that a general method to achieve this for any subpixel arrangement, can be incorporated in an image scaler at low additional cost, allowing simple quality trade-off control.

Complexity scalable motion estimation

Complexity scalable algorithms are capable of trading resource usage for output quality in a near-optimal way. We present a complexity scalable motion estimation algorithm based on the 3-D recursive search block matcher. We introduce data prioritizing as a new approach to scalability. With this approach, we achieve a near-constant complexity and a continuous quality-resource distribution. While maintaining acceptable quality, it is possible to vary the resource usage from below 1 match-error calculation per block on the average to more than 5 match-error calculations per block on the average.

Quality adaptive least squares trained filters for video compression artifacts removal using a no-reference block visibility metric

Compression artifacts removal is a challenging problem because videos can be compressed at different qualities. In this paper, a least squares approach that is self-adaptive to the visual quality of the input sequence is proposed. For compression artifacts, the visual quality of an image is measured by a no-reference block visibility metric. According to the blockiness visibility of an input image, an appropriate set of filter coefficients that are trained beforehand is selected for optimally removing coding artifacts and reconstructing object details. The performance of the proposed algorithm is evaluated on a variety of sequences compressed at different qualities in comparison to several other de-blocking techniques. The proposed method outperforms the others significantly both objectively and subjectively.

Graceful Degradation in Motion-Compensated Field-Rate Conversion

Abstract Recent motion estimation algorithms have reached a quality level that allows an improved motion portrayal for field rate conversion systems. As, in practice, still situations may occur in which motion estimation fails, a strategy for graceful degradation is required, to prevent the possible artifacts resulting from the processing from outweighing its advantages. In this paper both a global fall back detection and processing mode is introduced, and a novel method applying ordered statistical filtering in the up-convertor that realizes a graceful degradation for local errors in the estimated motion vector field.

Video-based respiration monitoring with automatic region of interest detection

Vital signs monitoring is ubiquitous in clinical environments and emerging in home-based healthcare applications. Still, since current monitoring methods require uncomfortable sensors, respiration rate remains the least measured vital sign. In this paper, we propose a video-based respiration monitoring method that automatically detects a respiratory region of interest (RoI) and signal using a camera. Based on the observation that respiration induced chest/abdomen motion is an independent motion system in a video, our basic idea is to exploit the intrinsic properties of respiration to find the respiratory RoI and extract the respiratory signal via motion factorization. We created a benchmark dataset containing 148 video sequences obtained on adults under challenging conditions and also neonates in the neonatal intensive care unit (NICU). The measurements obtained by the proposed video respiration monitoring (VRM) method are not significantly different from the reference methods (guided breathing or contact-based ECG; p-value  =  0.6), and explain more than 99% of the variance of the reference values with low limits of agreement (-2.67 to 2.81 bpm). VRM seems to provide a valid solution to ECG in confined motion scenarios, though precision may be reduced for neonates. More studies are needed to validate VRM under challenging recording conditions, including upper-body motion types.

Motion robust PPG-imaging through color channel mapping

Photoplethysmography (PPG)-imaging is an emerging noninvasive technique that maps spatial blood-volume variations in living tissue with a video camera. In this paper, we clarify how cardiac-related (i.e., ballistocardiographic; BCG) artifacts occur in this imaging modality and address these using algorithms from the remote-PPG literature. Performance is assessed under stationary conditions at the immobilized hand. Our proposal outperforms the state-of-the-art, blood pulsation imaging [Biomed. Opt. Express 5, 3123 (2014)25401026. ], even in our best attempt to create diffused illumination. BCG-artifacts are suppressed to an order of magnitude below PPG-signal strength, which is sufficient to prevent interpretation errors.

Comparison of machine learning techniques for target detection

This paper focuses on machine learning techniques for real-time detection. Although many supervised learning techniques have been described in the literature, no technique always performs best. Several comparative studies are available, but have not always been performed carefully, leading to invalid conclusions. Since benchmarking all techniques is a tremendous task, literature has been used to limit the available options, selecting the two most promising techniques (AdaBoost and SVM), out of 11 different Machine Learning techniques. Based on a thorough comparison using 2 datasets and simulating noise in the feature set as well as in the labeling, AdaBoost is concluded to be the best machine learning technique for real-time target detection as its performance is comparable to SVM, its detection time is one or multiple orders of magnitude faster, its inherent feature selection eliminates this as a separate task, while it is more straightforward to use (only three coupled parameters to tune) and has a lower training time.