Resolving overlapping convex objects in silhouette images by concavity analysis and Gaussian process

Zafari, Sahar; Murashkina, Mariia; Eerola, Tuomas; Sampo, Jouni; Kälviäinen, Heikki; Haario, Heikki

doi:10.1016/j.jvcir.2020.102962

Cited by 12 publications

(3 citation statements)

References 43 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…However, the algorithm is not capable of detecting primary particles within agglomerates in the case of particle breakage. In addition, although convolutional neural network (CNN)-based methods have significantly contributed to the detection and quantitative analysis of agglomerates, they require extensive training sets to achieve high performance, and manually labeling contours is time-consuming and expensive …”

Section: Introductionmentioning

confidence: 99%

Synthetic Image Analysis for Accurate Agglomerate Characterization and Control in Liquid–Liquid Phase Separation Systems

Zhou,

Xuanyuan,

et al. 2024

Crystal Growth & Design

View full text Add to dashboard Cite

Agglomeration control is important for both the crystallization process and product quality optimization. Fast and precise description of the agglomerate is crucial for agglomeration control. This work proposed a synthetic image analysis method to accurately count primary particles within agglomerates by recognizing and clustering the salient features. In addition, a new definition of the agglomeration degree distribution was introduced to underscore the proportion of heavy agglomerates within the particle population. Further, the image analysis method and agglomeration degree distribution were tested in raspberry ketone-1-propanol-water system, which is a typical antisolvent oiling-out system accompanied by liquid–liquid phase separation (LLPS) and agglomeration. A seeding strategy was built up by investigating the effects of seeding conditions, including seed size and addition time, on agglomeration. The results demonstrated the effectiveness of the proposed method, which correctly counts more than 78% of the particles even for heavy agglomerates. The agglomeration was effectively inhibited via introducing small seeds after LLPS. The image analysis method and the concept of agglomeration degree distribution proposed in this work could be used to provide valuable insights into the agglomeration behavior and agglomeration control of similar systems.

show abstract

Section: Introductionmentioning

confidence: 99%

Synthetic Image Analysis for Accurate Agglomerate Characterization and Control in Liquid–Liquid Phase Separation Systems

Zhou,

Xuanyuan,

et al. 2024

Crystal Growth & Design

View full text Add to dashboard Cite

show abstract

“…Concavity measurement has been extensively studied in computational graphics [9,10]. It can be used for morphology decomposition [11][12][13], computation of shape similarity, and object indexing [14,15]. However, current concavity measurements for the concave particle morphology either lack a description of its concave features or are computationally expensive.…”

Section: Introductionmentioning

confidence: 99%

Morphological characterization of concave particle based on convex decomposition

Du,

Li,

Liu

et al. 2024

Meas. Sci. Technol.

View full text Add to dashboard Cite

Particle morphology is an important factor affecting the mechanical properties of granular materials. However, it is difficult to quantify the morphology characteristics of the complex concave particle. Fortunately, complex particle can be segmented by convex decomposition, so a new shape index named convex decomposition coefficient (CDC) related to the number of segmentations is proposed. First, the pocket concavity was introduced to simplify the morphology hierarchically. Second, the cut weight linked to concavity was defined and convex decomposition was linearly optimised by maximizing the total cut weights. Third, the CDC was defined as the minimum block number where the block area ratio cumulatively exceeded 0.9 in descending order. Finally, the proposed index was used to quantify the particle morphology of coral sand. The results demonstrate that the CDC of coral sands mainly ranges from 2 to 6, with a positively skewed distribution. Furthermore, CDC correlates well with three shape indices: sphericity, particle size, and convexity. Larger CDC is associated with smaller sphericity, larger particle size, and smaller convexity. The index has certain scientific research value and practical significance.

show abstract

“…The binarization is formulated as a semantic segmentation task in which each pixel in the image is assigned to an object class, the foreground and the background, and is implemented based on the well-known U-Net architecture [17]. We integrate the U-Net based binarization to our segmentation framework for overlapping NPs [24,25] that utilize concave points in the contours of nanoparticle bundles. We demonstrate that the U-Net based binarization produces accurate separation of NPs and background without causing fluctuations to the nanoparticle contours.…”

Section: Introductionmentioning

confidence: 99%

Automated Segmentation of Nanoparticles in BF TEM Images by U-Net Binarization and Branch and Bound

Zafari

Eerola

Ferreira

et al. 2019

Computer Analysis of Images and Patterns

Self Cite

View full text Add to dashboard Cite

Transmission electron microscopy (TEM) provides information about Inorganic nanoparticles that no other method is able to deliver. Yet, a major task when studying Inorganic nanoparticles using TEM is the automated analysis of the images, i.e. segmentation of individual nanoparticles. The current state-ofthe-art methods generally rely on binarization routines that require parameterization, and on methods to segment the overlapping nanoparticles (NPs) using highly idealized nanoparticle shape models. It is unclear, however, that there is any way to determine the best set of parameters providing an optimal segmentation, given the great diversity of NPs characteristics, such as shape and size, that may be encountered. Towards remedying these barriers, this paper introduces a method for segmentation of NPs in Bright Field (BF) TEM images. The proposed method involves three main steps: binarization, contour evidence extraction, and contour estimation. For the binarization, a model based on the U-Net architecture is trained to convert an input image into its binarized version. The contour evidence extraction starts by recovering contour segments from a binarized image using concave contour points detection. The contour segments which belong to the same nanoparticle are grouped in the segment grouping step. The grouping is formulated as a combinatorial optimization problem and solved using the well-known branch and bound algorithm. Finally, the full contours of the NPs are estimated by an ellipse. The experiments on a real-world dataset consisting of 150 BF TEM images containing approximately 2,700 NPs show that the proposed method outperforms five current state-of-art approaches in the overlapping NPs segmentation.

show abstract

Resolving overlapping convex objects in silhouette images by concavity analysis and Gaussian process

Cited by 12 publications

References 43 publications

Synthetic Image Analysis for Accurate Agglomerate Characterization and Control in Liquid–Liquid Phase Separation Systems

Synthetic Image Analysis for Accurate Agglomerate Characterization and Control in Liquid–Liquid Phase Separation Systems

Morphological characterization of concave particle based on convex decomposition

Automated Segmentation of Nanoparticles in BF TEM Images by U-Net Binarization and Branch and Bound

Contact Info

Product

Resources

About