.png){kind=icon}
Return
Vehicle type classification using graph ant colony optimizer based stack autoencoder model
.png){kind=icon}
PRE
AI
DOI：10.1007/s11042-021-11508-5.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
In the intelligent transport system, vehicle type classification technology plays a major role. With the growth of video processing and pattern recognition application, a deep learning model is proposed in this research article to improve vehicle type classification under dynamic background. Initially, the original video sequences are collected from MIOvision Traffic Camera Dataset (MIO-TCD), and CDnet2014 dataset. Additionally, the contrast and visible level of the video frames are improved by implementing histogram equalization method. Next, the moving vehicles are detected and tracked using Gaussian Mixture Model (GMM) and Kalman filter. Then, the feature extraction is accomplished using Dual Tree Complex Wavelet Transform (DTCWT), Histogram of Oriented Gradients (HOG), and Local Ternary Pattern (LTP) to extract the texture feature vectors. Further, a new graph clustering-Ant Colony Optimization (ACO) algorithm is proposed to select the active feature vectors for better vehicle type classification. Lastly, the selected active feature vectors are given as the input to stack autoencoder classifier to classify eleven vehicle types in MIO-TCD and four vehicle types in CDnet2014 dataset. In the experimental section, the graph ACO based stack autoencoder model achieved 99.09%, and 89.89% of classification accuracy on both MIO-TCD, and CDnet2014 dataset, which are better related to the existing models like attention based method, improved spatiotemporal sample consistency algorithm, and generative adversarial nets.
Keywords:
Ant colony optimizer
Gaussian mixture model
Histogram equalization
Histogram of oriented gradients
Local ternary pattern
Stack autoencoder
Vehicle type classification
Journal
IF:
3
Papers:
2.0W
Citations:
3.2W
Organization
No organization information available
Cited Papers
.png){kind=icon}
ShareSaveShareSave
ShareSaveShareSave
View more