Accurate, real-time, unadorned lip tracking

Kaucic, R.; Blake, A.

doi:10.1109/iccv.1998.710745

Cited by 39 publications

(25 citation statements)

References 10 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Adaptive thresholds representation as an operation of the grey scale histogram of the image were then employed to segment the lip pixels. [59] used RGB values from training images to learn the Fisher discriminant axis. The mouth region colour onto their axes used to enhance the lip-skin boundary then a threshold was applied to segment the lip pixels.…”

Section: Related Workmentioning

confidence: 99%

4D Automatic Lip-Reading for Speaker's Face Identification

AbUShana¹

2018

SIPIJ

View full text Add to dashboard Cite

show abstract

Section: Related Workmentioning

confidence: 99%

4D Automatic Lip-Reading for Speaker's Face Identification

AbUShana¹

2018

SIPIJ

View full text Add to dashboard Cite

show abstract

“…The first method is gradientbased line search boundary detection [2] which looks for the first significant change in the intensity gradient along the searchline. The second method models the texture on each side of the boundary by a Gaussian distribution and uses Fisher's discriminant metric [17] to maximize the ratio of intraclass to interclass covariances. The third method is the Bayesian boundary estimation [5], which uses a uniform prior and a non-parametric texture process model on both sides of the boundary to maximize the probability of texture transition in a Bayesian sense using a small amount of observations.…”

Section: Evaluation and Comparisonmentioning

confidence: 99%

Classification-Based Probabilistic Modeling of Texture Transition for Fast Line Search Tracking and Delineation

Shahrokni

Drummond

Fleuret

et al. 2009

IEEE Trans. Pattern Anal. Mach. Intell.

View full text Add to dashboard Cite

Abstract-We introduce a classification-based approach to finding occluding texture boundaries. The classifier is composed of a set of weak learners which operate on image intensity discriminative features which are defined on small patches and fast to compute. A database which is designed to simulate digitized occluding contours of textured objects in natural images is used to train the weak learners. The trained classifier score is then used to obtain a probabilistic model for the presence of texture transitions which can readily be used for line search texture boundary detection in the direction normal to an initial boundary estimate. This method is fast and therefore suitable for real-time and interactive applications. It works as a robust estimator which requires a ribbon like search region and can handle complex texture structures without requiring a large number of observations. We demonstrate results both in the context of interactive 2-D delineation and fast 3-D tracking and compare its performance with other existing methods for line search boundary detection.

show abstract

“…Kalman filters [2], [3] and skin-color detection [1], [4], [5] have been commonly applied in face tracking as prediction filters. In our case, we have proved that integral projections can be used to solve the problem by themselves.…”

Section: Preprocessing Stepmentioning

confidence: 99%

“…Most of them are based on skin-like color detection [1]- [8], which is a well-known robust method to track human heads and hands. Color detection is usually followed by a location of individual facial features, for example, using PCA [2], splines [3] or integral projections [4], [5], [6]. However, in most of the existing research, projections are processed in a rather heuristic way.…”

Section: Introduction and Related Researchmentioning

confidence: 99%

Refining Face Tracking with Integral Projections

Garcı́a-Mateos

2003

Lecture Notes in Computer Science

View full text Add to dashboard Cite

Abstract. Integral projections can be used, by themselves, to accurately track human faces in video sequences. Using projections, the tracking problem is effectively separated into the vertical, horizontal and rotational dimensions. Each of these parts is solved, basically, through the alignment of a projection signal -a one-dimensional pattern-with a projection model. The effect of this separation is an important improvement in feature location accuracy and computational efficiency. A comparison has been done with respect to the CamShift algorithm. Our experiments have also shown a high robustness of the method to 3D pose, facial expression, lighting conditions, partial occlusion, and facial features.

show abstract

Accurate, real-time, unadorned lip tracking

Cited by 39 publications

References 10 publications

4D Automatic Lip-Reading for Speaker's Face Identification

4D Automatic Lip-Reading for Speaker's Face Identification

Classification-Based Probabilistic Modeling of Texture Transition for Fast Line Search Tracking and Delineation

Refining Face Tracking with Integral Projections

Contact Info

Product

Resources

About