This paper presents a tracking method based on parameters between colour blobs. The colour blobs
are obtained from segmenting the overall target into multiple colour regions. The colour regions are
segmented using EM method that determines the normal colour distributions from the overall colour
pixel distribution. After segmenting into different regions on the different colour layers, parameters
can be generated between colour regions of interest. In this instance, the colour regions of interest are
the top and bottom colour regions. The parameters that are generated from these colour regions are
the vector magnitude, vector angle and the value difference between colour regions. These parameters
are used as a means for tracking targets of interest. These parameters are used for tracking the target
of interest across an array of cameras which in this instance are three cameras. Three cameras have
been set up with different background and foreground conditions. The summarised results of tracking
targets across three cameras have shown that the consistency of colour regions across different cameras
and different background settings provided sufficient parameters for targets to be tracked consistently.
Example of tracking performance across three cameras were 0.88, 0.67 and 0.55. The remaining tracking
performances across three cameras are shown in Table 2. The tracking performance indicate that the
parameters between colour regions were able to be used for tracking a target across different cameras
with different background scenarios. Based on results obtained, parameters between segmented colour
regions have indicated robustness in tracking target of interest across three cameras.
The state of activated sludge wastewater treatment process (AS WWTP) is conventionally identified by physico-chemical measurements which are costly, time-consuming and have associated environmental hazards. Image processing and analysis-based linear regression modeling has been used to monitor the AS WWTP. But it is plant- and state-specific in the sense that it cannot be generalized to multiple plants and states. Generalized classification modeling for state identification is the main objective of this work. By generalized classification, we mean that the identification model does not require any prior information about the state of the plant, and the resultant identification is valid for any plant in any state. In this paper, the generalized classification model for the AS process is proposed based on features extracted using morphological parameters of flocs. The images of the AS samples, collected from aeration tanks of nine plants, are acquired through bright-field microscopy. Feature-selection is performed in context of classification using sequential feature selection and least absolute shrinkage and selection operator. A support vector machine (SVM)-based state identification strategy was proposed with a new agreement solver module for imbalanced data of the states of AS plants. The classification results were compared with state-of-the-art multiclass SVMs (one-vs.-one and one-vs.-all), and ensemble classifiers using the performance metrics: accuracy, recall, specificity, precision, F measure and kappa coefficient (κ). The proposed strategy exhibits better results by identification of different states of different plants with accuracy 0.9423, and κ 0.6681 for the minority class data of bulking.
Emotion is a crucial aspect of human health, and emotion recognition systems serve important roles in the development of neurofeedback applications. Most of the emotion recognition methods proposed in previous research take predefined EEG features as input to the classification algorithms. This paper investigates the less studied method of using plain EEG signals as the classifier input, with the residual networks (ResNet) as the classifier of interest. ResNet having excelled in the automated hierarchical feature extraction in raw data domains with vast number of samples (e.g., image processing) is potentially promising in the future as the amount of publicly available EEG databases has been increasing. Architecture of the original ResNet designed for image processing is restructured for optimal performance on EEG signals. The arrangement of convolutional kernel dimension is demonstrated to largely affect the model's performance on EEG signal processing. The study is conducted on the Shanghai Jiao Tong University Emotion EEG Dataset (SEED), with our proposed ResNet18 architecture achieving 93.42% accuracy on the 3-class emotion classification, compared to the original ResNet18 at 87.06% accuracy. Our proposed ResNet18 architecture has also achieved a model parameter reduction of 52.22% from the original ResNet18. We have also compared the importance of different subsets of EEG channels from a total of 62 channels for emotion recognition. The channels placed near the anterior pole of the temporal lobes appeared to be most emotionally relevant. This agrees with the location of emotion-processing brain structures like the insular cortex and amygdala.