Dmitry V. Yurin

Efficient sampling strategy and refinement strategy for randomized circle detection

Circle detection is fundamental in pattern recognition and computer vision. The randomized approach has received much attention for its computational benefit when compared with the Hough transform. In this paper, a multiple-evidence-based sampling strategy is proposed to speed up the randomized approach. Next, an efficient refinement strategy is proposed to improve the accuracy. Based on different kinds of ten test images, experimental results demonstrate the computation-saving and accuracy effects when plugging the proposed strategies into three existing circle detection methods.

Scale-space color blob and ridge detection

Feature detection in color images frequently consists in image conversion from color to grayscale and then one of grayscale detectors application. This approach has a few disadvantages: some features become indistinguishable in grayscale and features ordering based on response of grayscale detector do not accord with features order of importance from human’s perception point of view. There are two essential contributions in this paper. First, the method for direct detection of blobs and ridges in color images is proposed. Second, for scale-space ridge detection we introduce a 3D non maxima suppression procedure (in two orthogonal directions) which makes ridge detection simple and easy programmable in contrast to Lindeberg’s automatic scale selection approach. The proposed algorithms also produce estimates of blobs sizes and ridges width.

Preliminary coarse image registration by using straight lines found on them for constructing super resolution mosaics and 3D scene recovery

An algorithm of coarse image registration of a 3D scene taken from different camera perspectives is proposed. The algorithm uses information on geometrical parameters of straight lines found on the images and on distribution of color and/or brightness around these lines. Colors are taken into account by using the fuzzy logic technique. The result of the algorithm operation is a planar projective transformation (planar homography) matching approximately the images. In order to use the technique in algorithms of 3D scene reconstruction, an estimate of size of the window used for searching correspondent points after the coarse image registration is obtained.

HeNLM-LA: a locally adaptive non-local means algorithm based on hermite functions expansion

A new noise reduction algorithm HeNLM-LA is proposed. It is a modification of the non-local means algorithm using Hermite functions expansion of pixel neighborhoods. The filtering strength parameter is automatically adjusted proportionally to the local noise level. An algorithm for local noise level estimation is based on edge modeling; it suppresses high-amplitude edges in the map of local image variance.

Fast parametric curves detection based on statistical hypotheses estimation

We propose a new fast and reliable algorithm of parametric curves detection on images. In our approach as well as in many other approaches based on the Hough transform, we analyze the set of points obtained by an edge detector. Edge points are collected into chains of connected pixels, and then they are analyzed as a whole and piecemeal. At first we use randomized methods to reject unsuitable models. It gives us a high performance. Model parameters of fragments remained after randomized methods are estimated by the least squares method and the reliability of the hypothesis is estimated by the chi-square criterion. Then if the chi square criterion shows higher reliability for the merged similar models, we merge the obtained models.

Scale-space line curvature estimation for straight line and circle detection

A scale-space algorithm for estimating the local curvature of lines (edges or isolines) is presented. Two variants of edge curvature estimation based on differential invariants are suggested and compared. The first variant uses the edge curvature formula derived in the paper and the second is based on image preprocessing, which allows one to use the isoline curvature formula for edge curvature estimation. An analysis of scale selection needed to reach a desired accuracy is presented. Also, noise influence analysis has been performed. The application of curvature estimation to detection of straight lines and circles is suggested and implemented. Curvature information usage in parametric curve detection speeds up search algorithms and makes the results more stable.

Color Blob and line detection in scale-space

A scale-space algorithm for blobs and ridges detection in color images is proposed.

A Priori and A Posteriori Error Estimates in Recovery of 3D Scenes by Factorization Algorithms

Many various algorithms for recovering three-dimensional scenes from a set of digital images have currently been developed. For certain scenes and shooting conditions, some algorithms give nice results, whereas others produce unacceptable results. In this paper, for a group of algorithms based on the matrix factorization, criteria are derived that make it possible, (a) by known statistical characteristics of the scene and shooting conditions, to predict whether a given algorithm can be used in the given case and, if it can, to determine the expected accuracy, (b) when recovering an unknown scene, to compute not only the desired results but also their accuracy (authenticity). A modification of the algorithm based on the adaptive selection of the approximation is suggested. Experimental verification of the criteria and estimates obtained showed their high efficiency and reliability.

Circle detection based on multiscale edge curvature estimation

A procedure for adapting the isoline curvature formula to estimate an edge curvature in the images is presented. The curvature estimation is used in the framework of the scale-space algorithm, which makes it possible to choose the required scale adaptively at each point of interest. The procedure for edge points clustering to the subsets, each of which contains the points of one circumference, has been developed to detect the circumferences.

Image Ridge Denoising Using No-Reference Metric

Image denoising methods depend on inner parameters that control filter strength, so the problem of the filter parameters choice arises. Parameter optimization can be done in the ridge areas, when we can analyze their appearance on the difference between original noisy and filtered image (so-called method noise image). If this difference is irregular, then the filtering strength can be increased. If regular components appear on method noise, then the filtering strength is too large. We use mutual information closely connected with conditional entropy for the analysis and consider images corrupted with Gaussian-like noise with small correlation radius. Ridge detection approach based on Hessian matrix eigenvalues analysis is used for estimation of sizes and directions of image characteristic details. Retinal images containing many ridges of different scales and directions from DRIVE and general images from TID2008 databases with added controlled Gaussian noise were used for testing with NLM and LJNLM-LR methods.