This paper presents a method for extracting distinctive invariant features from images that can be used to perform reliable matching between different views of an object or scene. The features are invariant to image scale and rotation, and are shown to provide robust matching across a substantial range of affine distortion, change in 3D viewpoint, addition of noise, and change in illumination. The features are highly distinctive, in the sense that a single feature can be correctly matched with high probability against a large database of features from many images. This paper also describes an approach to using these features for object recognition. The recognition proceeds by matching individual features to a database of features from known objects using a fast nearest-neighbor algorithm, followed by a Hough transform to identify clusters belonging to a single object, and finally performing verification through least-squares solution for consistent pose parameters. This approach to recognition can robustly identify objects among clutter and occlusion while achieving near real-time performance.

/pdf/distinctive-image-features-from-scale-invariant-keypoints-3waurjqzke.pdf

Distinctive Image Features from Scale-Invariant Keypoints

We propose a deep convolutional neural network architecture codenamed Inception that achieves the new state of the art for classification and detection in the ImageNet Large-Scale Visual Recognition Challenge 2014 (ILSVRC14). The main hallmark of this architecture is the improved utilization of the computing resources inside the network. By a carefully crafted design, we increased the depth and width of the network while keeping the computational budget constant. To optimize quality, the architectural decisions were based on the Hebbian principle and the intuition of multi-scale processing. One particular incarnation used in our submission for ILSVRC14 is called GoogLeNet, a 22 layers deep network, the quality of which is assessed in the context of classification and detection.

/pdf/going-deeper-with-convolutions-1yobw2o2ds.pdf

Going deeper with convolutions

The Scale-Invariant Feature Transform (or SIFT) algorithm is a highly robust method to extract and consequently match distinctive invariant features from images. These features can then be used to reliably match objects in diering images. The algorithm was rst proposed by Lowe [12] and further developed to increase performance resulting in the classic paper [13] that served as foundation for SIFT which has played an important role in robotic and machine vision in the past decade.

In this paper, we present a novel scale- and rotation-invariant interest point detector and descriptor, coined SURF (Speeded Up Robust Features). It approximates or even outperforms previously proposed schemes with respect to repeatability, distinctiveness, and robustness, yet can be computed and compared much faster.

This is achieved by relying on integral images for image convolutions; by building on the strengths of the leading existing detectors and descriptors (in casu, using a Hessian matrix-based measure for the detector, and a distribution-based descriptor); and by simplifying these methods to the essential. This leads to a combination of novel detection, description, and matching steps. The paper presents experimental results on a standard evaluation set, as well as on imagery obtained in the context of a real-life object recognition application. Both show SURF's strong performance.

/pdf/surf-speeded-up-robust-features-1fje09jteo.pdf

SURF: speeded up robust features

https://www.cs.jhu.edu/~misha/ReadingSeminar/Papers/Bay08.pdf

Speeded-Up Robust Features (SURF)

This paper describes an approach for tracking rigid and articulated objects using a view-based representation. The approach builds on and extends work on eigenspace representations, robust estimation techniques, and parameterized optical flow estimation. First, we note that the least-squares image reconstruction of standard eigenspace techniques has a number of problems and we reformulate the reconstruction problem as one of robust estimation. Second we define a “subspace constancy assumption” that allows us to exploit techniques for parameterized optical flow estimation to simultaneously solve for the view of an object and the affine transformation between the eigenspace and the image. To account for large affine transformations between the eigenspace and the image we define a multi-scale eigenspace representation and a coarse-to-fine matching strategy. Finally, we use these techniques to track objects over long image sequences in which the objects simultaneously undergo both affine image motions and changes of view. In particular we use this “EigenTracking” technique to track and recognize the gestures of a moving hand.

EigenTracking: Robust Matching and Tracking of Articulated Objects Using a View-Based Representation

We propose a framework for learning robust, adaptive, appearance models to be used for motion-based tracking of natural objects. The model adapts to slowly changing appearance, and it maintains a natural measure of the stability of the observed image structure during tracking. By identifying stable properties of appearance, we can weight them more heavily for motion estimation, while less stable properties can be proportionately downweighted. The appearance model involves a mixture of stable image structure, learned over long time courses, along with two-frame motion information and an outlier process. An online EM-algorithm is used to adapt the appearance model parameters over time. An implementation of this approach is developed for an appearance model based on the filter responses from a steerable pyramid. This model is used in a motion-based tracking algorithm to provide robustness in the face of image outliers, such as those caused by occlusions, while adapting to natural changes in appearance such as those due to facial expressions or variations in 3D pose.

/pdf/robust-online-appearance-models-for-visual-tracking-509edop7ga.pdf

Robust online appearance models for visual tracking

We present a technique for the computation of 2D component velocity from image sequences. Initially, the image sequence is represented by a family of spatiotemporal velocity-tuned linear filters. Component velocity, computed from spatiotemporal responses of identically tuned filters, is expressed in terms of the local first-order behavior of surfaces of constant phase. Justification for this definition is discussed from the perspectives of both 2D image translation and deviations from translation that are typical in perspective projections of 3D scenes. The resulting technique is predominantly linear, efficient, and suitable for parallel processing. Moreover, it is local in space-time, robust with respect to noise, and permits multiple estimates within a single neighborhood. Promising quantiative results are reported from experiments with realistic image sequences, including cases with sizeable perspective deformation.

Computation of component image velocity from local phase information

As an observer moves and explores the environment, the visual stimulation in his/her eye is constantly changing. Somehow he/she is able to perceive the spatial layout of the scene, and to discern his/her movement through space. Computational vision researchers have been trying to solve this problem for a number of years with only limited success. It is a difficult problem to solve because the optical flow field is nonlinearly related to the 3D motion and depth parameters. Here, we show that the nonlinear equation describing the optical flow field can be split by an exact algebraic manipulation to form three sets of equations. The first set relates the flow field to only the translational component of 3D motion. Thus, depth and rotation need not be known or estimated prior to solving for translation. Once the translation has been recovered, the second set of equations can be used to solve for rotation. Finally, depth can be estimated with the third set of equations, given the recovered translation and rotation. The algorithm applies to the general case of arbitrary motion with respect to an arbitrary scene. It is simple to compute, and it is plausible biologically. The results reported in this article demonstrate the potential of our new approach, and show that it performs favorably when compared with two other well-known algorithms.

Subspace methods for recovering rigid motion I: algorithm and implementation

The measurement of image disparity is a fundamental precursor to binocular depth estimation. Recently, Jenkin and Jepson (in Computational Processes in Human Vision (V. Pylyshyn, Ed.), Ablex, New Jersey, 1988) and Sanger (Biol. Cybernet, 59, 1988 , 405–418) described promising methods based on the output phase behavior of bandpass Gabor filters. Here we discuss further justification for such techniques based on the stability of bandpass phase behavior as a function of typical distortions that exist between left and right views. In addition, despite this general stability, we show that phase signals are occasionally very sensitive to spatial position and to variations in scale, in which cases incorrect measurements occur. We find that the primary cause for this instability is the existence of singularities in phase signals. With the aid of the local frequency of the filter output (provided by the phase derivative) and the local amplitude information, the regions of phase instability near the singularities are detected so that potentially incorrect measurements can be identified. In addition, we show how the local frequency can be used away from the singularity neighbourhoods to improve the accuracy of the disparity estimates. Some experimental results are reported.

Allan D. Jepson

Papers

EigenTracking: Robust Matching and Tracking of Articulated Objects Using a View-Based Representation

Robust online appearance models for visual tracking

Computation of component image velocity from local phase information

Subspace methods for recovering rigid motion I: algorithm and implementation

Phase-based disparity measurement