Mingli Lu

Modified unsupervised discriminant projection with an electronic nose for the rapid determination of Chinese mitten crab freshness

In this paper, a method to rapidly determine living Chinese mitten crab freshness using an electronic nose (E-nose) and a non-linear data processing technique was studied. E-nose responses to crabs stored at 4 °C were measured. Meanwhile, total volatile basic nitrogen (TVB-N) was examined to provide freshness references for the E-nose analysis. However, traditional classification algorithms are not suitable for E-nose data with non-linear manifold structures; therefore, a modified unsupervised discriminant projection (MUDP) coupled with sample label information was proposed. MUDP can retain the local and global structure and take advantage of the importance of label information and then create a geometric structure optimal linear projection. Data classification experimental results proved that the classification accuracy of K-nearest-neighbor (KNN) combined with the data processed by MUDP was much better than that of other considered methods. Validation experiments indicated that the recognition rates of the proposed algorithm were higher than those of traditional linear algorithms such as PCA and LDA or nonlinear algorithms such as KPCA.

A multiCell visual tracking algorithm using multi-task particle swarm optimization for low-contrast image sequences

In terms of the varying number of cell population, shape deformation, collision and uneven movement, a novel method based on multi-task particle swarm optimization (PSO) algorithm without explicit detection module, named MTPSO tracking method, is developed for automatic tracking of biological cells in time-lapse low-contrast microscopy image sequences. For tracking existing cells from the previous frames, a PSO-based tracking module is firstly implemented to give the initial positions of existing cells according to the previous estimated state of each cell, then a PSO-based contour module is proposed to determine the corresponding contour of each cell and finally achieve a precise position tracking by an iterative centroid updating process. For tracking new appearing cells at the current frame, a PSO-based discovery module, followed by the aforementioned PSO-based contour module, is proposed to search for new potential cells through appropriate initialization of particle swarm and searching mechanism. MTPSO tracking method is tested over a number of different real cell image sequences and is shown to provide high accuracy both in position and contour estimate of each cell in various challenging cases. Furthermore, it is more competitive against the state-of-the-art multi-object tracking methods in terms of performance measures such as FAR, FNR, LTR, and LSR.

Multiple cells tracking by firework algorithm

Multiple cells dynamics analysis through fluorescence microscopy imaging requires simultaneously tracking large and time-varying number cells in noisy image sequences. Such process is characterized as a challenging task due to several roadblocks including the severe image noise and clutter, the occlusion of one cell by others, and the weak image contrast. To gain full dynamics of multiple cells, a novel firework explosion searching behavior based tracking algorithm is proposed to tracking multiple cells in fluorescence image sequences in this paper. Each firework (spark) determines the state and then adjusts its following explosion according to cell detection position heuristic information. Simulation results verify the effectiveness of the methods.

A Cluster Based Method for Cell Segmentation

In the field of cell biology, cell segmentation is an essential task in biomedical application. For this purpose, a cluster based method for cell segmentation is proposed. Firstly, an ant colony clustering algorithm is used to make pre-segmentation from which cell candidates are identified, then some noise spots are filtered with area feature, after that, a novel cluster algorithm is proposed to divide adhering cells into individuals. Finally, good results of segmentation can be achieved. Experimental result show that the method remains both the advantage of image segment of ant colony cluster and the ability of further process of pre-segmentation, which improves the performance of cell segmentation.

Two Ant Decision Levels and Its Application to Multi-Cell Tracking

Inspired by ant’s stochastic behavior in searching of multiple food sources, a novel ant system with two ant decision levels are proposed to track multiple cells in biological field. In the ant individual level, ants within the same colony perform independently, and ant decision is determined in probability by both its intended motion model and likelihood function. In the ant cooperation level, each ant adjusts individual state within its influence region, while the global best template at current iteration is found among all ant colonies and further utilized to update ant model probability, influence region, and the probability of fulfilling task. Simulation results demonstrate that our algorithm could automatically track numerous cells and its performance is compared with the multi-Bernoulli filtering method.

An Ant–Based Filtering Random–Finite–Set Approach to Simultaneous Localization and Mapping

Abstract Inspired by ant foraging, as well as modeling of the feature map and measurements as random finite sets, a novel formulation in an ant colony framework is proposed to jointly estimate the map and the vehicle trajectory so as to solve a feature-based simultaneous localization and mapping (SLAM) problem. This so-called ant-PHD-SLAM algorithm allows decomposing the recursion for the joint map-trajectory posterior density into a jointly propagated posterior density of the vehicle trajectory and the posterior density of the feature map conditioned on the vehicle trajectory. More specifically, an ant-PHD filter is proposed to jointly estimate the number of map features and their locations, namely, using the powerful search ability and collective cooperation of ants to complete the PHD-SLAM filter time prediction and data update process. Meanwhile, a novel fast moving ant estimator (F-MAE) is utilized to estimate the maneuvering vehicle trajectory. Evaluation and comparison using several numerical examples show a performance improvement over recently reported approaches. Moreover, the experimental results based on the robot operation system (ROS) platform validate the consistency with the results obtained from numerical simulations.

A Novel Multi-cell Multi-Bernoulli Tracking Method Using Local Fractal Feature Estimation

A novel multi-cell tracking method based on multi-Bernoulli filter using local fractal feature estimation is proposed in this paper. The Hurst coefficient estimated by the rescaled range analysis method is considered as the local fractal feature. The local fractal feature can offer two advantages for multi-Bernoulli filter. The input of filter is the Hurst coefficient image, the direct effect is that observation can be modeled simply. And the likelihood function can be computed easily using this feature. Experiment results show that our proposed method could achieve an accurate and joint estimate of the number of cells and their individual states especially in the case of the number of cell population varying and the cellular morphology changing. And it shows equivalent accuracy against other tracking methods.

A Hybrid ACO-ACM Based Approach for Multi-cell Image Segmentation

In this paper, a hybrid multi-cell image segmentation approach is proposed, based on the combination of active contour model (ACM) and ant colony optimization (ACO), for multi-cell image segmentation. This novel image segmentation algorithm integrates the characteristics of ACM model into the ACO with tractable and well defined energy and heuristic functions. Consequently, the problem of cell image segmentation is actually converted to search for the marks of cell contours by group of ants. Experiment results show that our proposed approach is more effective than several existing methods, and it is noted that our proposed approach is developed and implemented in LabVIEW as well with performance consistency.

Research on Freshness Detection for Chinese Mitten Crab Based on Machine Olfaction

Aquatic products freshness detection is an important topic in the current issue of food quality and safety. In this paper, we presented an automatic device based on electronic nose for evaluation freshness of Chinese mitten crab. The crabs were stored at 4 °C for nine days. Electronic nose sensor responses of each sensor over the array were collected from the living crab samples in parallel with data from microbiological analysis for total volatile basic nitrogen (TVB-N). Qualitative interpretation of response data was based on sensory evaluation discriminating samples in three quality classes (fresh, semi-fresh, and spoiled). Principal component analysis (PCA), linear discriminant analysis (LDA), kernel principal component analysis (KPCA) and Laplacian Eigenmap (LE) were developed to classify crab samples in the respective quality class with response data. Experiment results indicated that LE outperform other methods and achieve the highest recognition accuracy for crabs with three quality classes.

Multi region segmentation algorithm based on edge preserving for molten pool image

This paper is aimed at the difficult problem of multi region segmentation of weld pool image, analyzed The difficulty of edge extraction in the inner region of the weld pool. According to the characteristics between pixel neighborhood space and neighbor pixel correlation, based on local standard deviation, presented a noise suppression, edge enhancement of the weld pool image multi region division and multi region edge detection algorithm, Through the test of the weld pool image, It shows that the algorithm can accurately divide the internal details of the weld pool. Finally, the Sobel operator, Roberts operator, Prewitt operator and the edge detection results of the weld pool image are analyzed and compared by experiments, The results show that the algorithm in this paper is much better than other algorithms, At last, the accuracy of the algorithm is tested by the difference shadow detection, a continuous multi region edge was obtained by the expansion of corrosion.