Recognition of overlapping elliptical objects in a binary image

Abstract

Recognition of overlapping objects is required in many applications in the field of computer vision. Examples include cell segmentation, bubble detection and bloodstain pattern analysis. This paper presents a method to identify overlapping objects by approximating them with ellipses. The method is intended to be applied to complex-shaped regions which are believed to be composed of one or more overlapping objects. The method has two primary steps. First, a pool of candidate ellipses are generated by applying the Euclidean distance transform on a compressed image and the pool is filtered by an overlaying method. Second, the concave points on the contour of the region of interest are extracted by polygon approximation to divide the contour into segments. Then, the optimal ellipses are selected from among the candidates by choosing a minimal subset that best fits the identified segments. We propose the use of the adjusted Rand index, commonly applied in clustering, to compare the fitting result with ground truth. Through a set of computational and optimization efficiencies, we are able to apply our approach in complex images comprised of a number of overlapped regions. Experimental results on a synthetic data set, two types of cell images and bloodstain patterns show superior accuracy and flexibility of our method in ellipse recognition, relative to other methods.

Influence of the hierarchical architecture of multi-core iron oxide nanoflowers on their magnetic properties

“…6d). Individual cores were localized using the Distance Transform-based Ellipse Contour Matching Algorithm (DTECMA) 60 that was applied to binary images of individual IONFs (Fig. 6e).…”

Unsupervised domain adaptation method for segmenting cross-sectional CCA images

“…The ellipse that is fit to this connected region therefore does not correspond well to the CCA. To prevent this, a region splitting algorithm inspired from [37] was implemented that aims to detect whether a connected component contains multiple ellipse centers, and splits the region in two if this is the case. First, in Fig.…”

Towards a Likelihood Ratio Approach for Bloodstain Pattern Analysis

“…A limitation of this method is that regionprops only fits each region with one ellipse while regions that may be composed of multiple droplets are discarded, which inevitably results in a loss of information. Instead, we use the algorithm developed in Zou et al (2021) that can approximate a non-elliptical region with a set of overlapping ellipses. The idea of the algorithm is first to generate a pool of candidate ellipses through a series of distance transforms of the region, then partition the region contour into several segments defined by its concave points, and finally match each segment with an ellipse from the candidate pool.…”

“…In terms of modeling the likelihood of a bloodstain pattern, our approach consists of three main steps: first, the bloodstain pattern image is processed and represented by a collection of ellipses approximating the constituent stains. We applied the technique from the work of Zou et al (2021) to infer the number of ellipses and their numerical characteristics of the ellipses. Second, a few predefined quantitative features are extracted from the ellipse representation of each pattern.…”