Intelligent traffic analysis: A heuristic high-dimensional image search algorithm based on spatiotemporal probability for constrained environments

The intent of this research is to design a license plate recognition (LPR) system in the domain of Bangla language for smart vehicle management. The proposed system is designed on the basis of computer vision tools and deep supervised machine learning model. The system has three modules: license plate detection, character segmentation and recognition of the characters of the License Plate (LP). The goal of detection is to localize the plate area from the vehicle image and to crop region of interest (LP). It is executed by applying following process: preprocessing the image, conversion to binary image, contour detection and filtering the contours to get the LP's character contours, tilt correction and cropping the plate area from the image. Then, the cropped LP is segmented to extract the characters from the plate. Finally, the recognition step classifies the characters by means of deep convolution neural network where the features of the character are crafted and learned by the convolution layers of the networks. The system is implemented in Python OpenCV environment for offline car license plates images which are taken in different illuminations, road scenarios and colored cars. The system performance is evaluated in terms of detection rate, segmentation rate, recognition rate and execution time. The results illustrate that the performance of the system is remarkable.

Smart cities must have all the important characteristics to achieve their intended goals. Proper traffic management and controlling, increased surveillance and safety, and enhanced management and avoidance of incidents must be the priorities of smart cities. Meanwhile, license plate recognition (LPR) has become the most debatable topic in the research community due to various real-time applications, such as “law enforcement, toll-free processing, access control, and traffic surveillance.” Automated LPR is a technique based on computer vision to recognize vehicles with their number plates. This study discusses various “Deep Learning based LPR” techniques to detect and identify “alphanumeric characters” in number plate. The projected model works on “license based detection” and “character recognition.” This technique uses “Optical Character Recognition (OCR)” technology to detect and extract the alphanumerical numbers from the license plate. This study is based on secondary data collected from various studies conducted on Licence Plate Recognition using various Deep Learning models published in databases like Google Scholar, Science Direct, NCBI, etc. Deep learning has been used widely in applications related to computer vision in recent years with great perfection. It is a great solution for modern and traditional image processing, object detection, and feature extraction issues. It has been widely used in different stages of LPR like character segmentation, license plate recognition, and OCR.

License plate location is the first and crucial stage in license plate recognition. Although many license plate location methods achieve higher success rate, they are always time-consuming. However, robust real-time characteristic is important to application of license plate recognition. In this paper, a texture-based method of license plate location is proposed to enhance real-time characteristic while keeping higher success rate. It consists of three parts: preprocessing, region of interested (ROI) location and ROI filtration. Experimental has demonstrated that the proposed method achieve average execution time of 49ms and success rate of 96.3% simultaneously.

A Real-World Size Organization of Object Responses in Occipitotemporal Cortex

Most of the existing redundant configuration methods of modular level converters (MMC) only consider the reliability of the MMC, and rarely consider the utilization of insulated gate bipolar transistors (IGBT). A heuristic algorithm-based calculation method for hybrid MMC redundancy configuration is proposed in order to better meet the engineering practicality. Firstly, the topology of the hybrid MMC is introduced, and the correlation between submodules is introduced into the calculation of its reliability. Secondly, the index of IGBT utilization is proposed to judge the economic advantage. The effects of correlation and service life on the reliability of the hybrid MMC and the IGBT utilization are analyzed. Then, heuristic factors are constructed with reliability and IGBT utilization, and the optimal redundancy configuration scheme of hybrid MMC is obtained by heuristic algorithm. Finally, based on the parameters of Kunliulong multi-terminal hybrid DC transmission project, the calculation is verified in Matlab and compared with the results of the difference method to illustrate the feasibility and superiority of the proposed redundancy configuration calculation method.

A comparative runtime analysis of heuristic algorithms for satisfiability problems

We construct an automatic and adaptive image analysis method for machine vision after the Marr primal sketch model for human visual perception. This method consists of four steps: image acquisition, object location, object validation, and parameter extraction coupled with object categorization. Any method that uses multiple threshold conditions can result in redundant objects that represent individual image features. Our method uses this redundancy to provide a best representation of each image feature. This object validation and categorization consists of the following steps: prescreening objects based on minimum validity criteria; constructing an object hierarchy based on containedness; reducing redundancy in this hierarchy but retaining multiple contained objects, thus performing a declumping process; identifying and separating heterogeneous objects from containing objects; and extracting parameters for final object categorization. We describe the application of our methods as an integrated image analysis solution to a variety of image processing problems. Qualitative and quantitative results are presented.

Higher order spatial features, such as doublets or triplets have been used to incorporate spatial information into the bag-of-local-features model. Due to computational limits, researchers have only been using features up to the 3rd order, i.e., triplets, since the number of features increases exponentially with the order. We propose an algorithm for identifying high-order spatial features efficiently. The algorithm directly evaluates the inner product of the feature vectors from two images to be compared, identifying all high-order features automatically. The algorithm hence serves as a kernel for any kernel-based learning algorithms. The algorithm is based on the idea that if a high-order spatial feature co-occurs in both images, the occurrence of the feature in one image would be a translation from the occurrence of the same feature in the other image. This enables us to compute the kernel in time that is linear to the number of local features in an image (same as the bag of local features approach), regardless of the order. Therefore, our algorithm does not limit the upper bound of the order as in previous work. The experiment results on the object categorization task show that high order features can be calculated efficiently and provide significant improvement in object categorization performance.

Operator context scanning to support high segmentation rates for real time license plate recognition

Recently, manipulation of real-world images has been highly elaborated along with the development of Generative Adversarial Networks (GANs) and corresponding encoders, which embed real-world images into the latent space. However, designing encoders of GAN still remains a challenging task due to the trade-off between distortion and perception. In this paper, we point out that the existing encoders try to lower the distortion not only on the interest region, e.g., human facial region but also on the uninterest region, e.g., background patterns and obstacles. However, most uninterest regions in real-world images are located at out-of-distribution (OOD), which are infeasible to be ideally reconstructed by generative models. Moreover, we empirically find that the uninterest region overlapped with the interest region can mangle the original feature of the interest region, e.g., a microphone overlapped with a facial region is inverted into the white beard. As a result, lowering the distortion of the whole image while maintaining the perceptual quality is very challenging. To overcome this trade-off, we propose a simple yet effective encoder training scheme, coined IntereStyle, which facilitates encoding by focusing on the interest region. IntereStyle steers the encoder to disentangle the encodings of the interest and uninterest regions. To this end, we filter the information of the uninterest region iteratively to regulate the negative impact of the uninterest region. We demonstrate that IntereStyle achieves both lower distortion and higher perceptual quality compared to the existing state-of-the-art encoders. Especially, our model robustly conserves features of the original images, which shows the robust image editing and style mixing results. We will release our code with the pre-trained model after the review.KeywordsStyleGANRobust GAN inversionInterest regionInterest disentanglementUninterest filter

In the previous chapters, we introduce how to improve the basic BBO algorithm by using local topologies and developing new migration operators. Besides improving the intrinsic structure and operators of the algorithm, another way to improve the algorithm is combining it with other heuristic algorithms. Based on its distinctive migration mechanisms, BBO has a good local exploitation ability (Simon, IEEE Trans. Evol. Comput. 12:702–713, 2008, [10]), but its global exploration ability is relatively poor. Thus, those hybrid BBO algorithms often introduce effective global exploration mechanisms of other heuristic algorithms, so as to better balance the global and local search. This chapter describes some typical hybrid BBO algorithms.

Combined Petri net modelling and AI-based heuristic hybrid search for flexible manufacturing systems—part II. Heuristic hybrid search

Optical Character Recognition (OCR) research includes computer vision, artificial intelligence, and pattern recognition. Character recognition has garnered a lot of attention in the last decade due to its broad variety of uses and applications, including multiple-choice test data, business documents (e.g., ID cards, bank notes, passports, etc.), and automatic number plate recognition. This paper introduces an automatic recognition system for printed numerals. The automatic reading system is based on extracting local statistical and geometrical features from the text image. Those features are represented by eight vectors extracted from each digit. Two of these features are local statistical (A, A th), and six are local geometrical (P1, P2, P3, P4, P5, and P6). Thus, the database created consists of 1120 statistical and geometrical features. For the purpose of recognition, the features of the test image are compared with the features of all the images saved in the database depending on the value of the Minimum Distance (MD). All digits (0–9) were identified with 100% accuracy. The average computational time required to recognize a numeral at any font size is 0.06879 seconds.

Trust propagation is being increasingly adopted to assist recommendation systems in providing more reliable information, upon which, users can make more accurate decisions. Optimal trust path search integrating trust value and path length plays a critical role in trust propagation, but suffers from insufficient performance regarding search accuracy and time. Generally, the quality of trust propagation is affected by the path length, and the longer the path is, the worse the trust quality is. However, the length of the path is not the unique crucial factor. Some longer paths with greater trust values may be more credible. In addition, the A* algorithm can find an optimal solution, but it expends much time to distinguish some similar paths. The A* algorithm is improved and a dynamic weighted heuristic trust path search algorithm is proposed. According to the six-degree space theory, the paths are extended to six-degree admissible trust paths. Then, according to the depths of the nodes in the search path, it relaxes the evaluation functionf(n) by devising a dynamic weighted factor w, inserts all nodes satisfied specific conditions into the FOCAL list. Furthermore, it sets the secondary heuristic factor and selects the nodes with the minimum heuristic factor value to reach the target node, and outputs the optimal trust path. Experiments on the public Advogato and FilmTrust datasets demonstrated that the proposed algorithm could efficiently identify the reliable trust paths and predict trust value with high accuracy and reduced computational complexity. The proposed algorithm could be applied to recommendation systems in the future.

In this work, we present a C++ implementation of object categorization with the bag-of-word (BoW) framework. Unlike typical BoW models which consider the whole area of an image as the region of interest (ROI) for visual codebook generation, our implementation only considers the regions of target objects as ROIs and the unrelated backgrounds will be excluded for generating codebook. This is achieved by a supervised mean shift algorithm. Our work is on the benchmark SIVAL dataset and utilizes a Maximum Margin Supervised Topic Model for classification. The final performance of our work is quite encouraging.