Embedded automatic vehicle identification system
Engineering, etc.
(Research goals, research background and current situation, schedule of works and project ideas, etc.)< /p>
BR />See appendix.
1. The advancement of the project:
With the rapid development of digital information technology and network technology in the post-PC era, the performance of embedded processors and high-performance processors have been able to satisfy themselves. Evaluation of complex algorithms and other complex applications, as well as embedded applications will inevitably enter various fields. On the other hand, with the rapid economic development of China and the Beijing Olympics, "intelligent transportation will become an indisputable hot topic. Due to the particularity of the transportation industry, the technical parameters of its equipment, and the stringent requirements for use conditions, embedded technology can just To meet this requirement, the widespread application of embedded intelligent transportation equipment is an inevitable trend. As the embedded automatic vehicle identification system is an important part of the intelligent traffic management system, it is a perfect combination of embedded technology and vehicle identification technology. , including three main features: embedded license plate recognition, embedded car logo recognition and automatic color recognition, and strives to lock the car in one go
It has the following advantages:
1. Highly. Independence: The use of embedded technology is only connected through the communication interface and application system independence
2. Full-featured: One-time targeted identification of license plates, vehicle logos and colors. The existing system has powerful functions.
3. Plasticity: It can be combined with the built-in wireless network and various serial interfaces of the upstream end of the front-end signal triggering device to combine with the downstream product system functions and scope of use. Extension.
4. Easy maintenance:
2. Operability and achievability:
Currently, you are waiting for the maturity and improvement of license plate recognition and vehicle recognition technology. Relevant information is easier to obtain. The existing embedded technology is relatively mature, so it is easier to implement than other cutting-edge scientific topics. The equipment and materials are also more accessible and the cost is moderate. p>
3. Innovation:
Existing license plate recognition equipment usually uses computers to process data, and some even require the cooperation of several computers, taking up a lot of space and resources, even if it is completed occasionally Embedded system, its function is limited to license plate recognition or identification of car logos. The system creatively combines embedded license plate recognition, vehicle logo recognition, and automatic color recognition, a one-time solution that makes system integration difficult for bloated equipment. , Poor stability, difficult to maintain, a problem with functionality.
4. Possible problems:
At present, the main problem is embedded integration and wireless transmission distance. Next, we imagine: Nowadays, most computers are used to process data, and the shortcomings of inflexible devices are to develop a portable wireless data transmission system that can automatically identify. However, due to our time, energy and money, "portability". "The degree of limitation is the biggest problem. In addition, the speed and depth of field image recognition problems are problems we may face.
Expected results
(Specific form of results achieved , such as: patents, published papers, and physical production techniques (including software programs), which can be the result of various forms)
We anticipate the results of our experiments.
First, we plan to build a complete embedded system technology that comes with tangible results.
Secondly, we analyze the market situation and the market prospects are very promising for the embedded vehicle identification system, which can be produced with our patent and market.
The third aspect, the color of the car, the license plate, the subject combination of the vehicle to determine the appropriate algorithm, so in the process of completing the system is inevitable, complete the design of the algorithm, this is part The results are published in the form of a paper performance.
Because we plan to complete the system, we need to complete both the hardware and software parts of the system.
From the point that a large part of the software and algorithm results can be published through newspapers and can be put into production and patented hardware results can be obtained to reflect that there is no doubt that our research results will not only be part of the software or only the hardware part theme, this is a big advantage.
The budget content, budget amount, and estimated execution time required for the experimental environment
Front-end image acquisition of CCD camera, purchase of camera or camera 3000 07.12 to 08.2
Special environmental light addition of auxiliary light source 1500 07.1208.2 months,
Analog signal digitization of image capture card 2500 07.1208.2 months
>Embedded system hardware Facilities, image processing 4000 08.3 08.10 to 08.12?08.10
DVR video information storage 2500
Display device output image recognition results 1500 08.1209.2 months
Wireless transceiver or cable transmission equipment information transmission 250 009 2?09.3
The final stage of processing mechanical parts and assembling them into prototype 2000
Total: 19,500 yuan
College Approval Opinions
Expert Committee’s Review Opinions
School Approval Opinions
Attachment: Current Status, Background and Significance of the Topic,
Since Since the birth of the world's first car in 1885, cars have had a huge impact on our daily work and lives. For more than a hundred years, cars with their low cost, easy operation, and advantages have gradually been accepted by the public, reaching hundreds of thousands of families. In China, many people join the car ownership group every year. Followed by the natural growth of fast and convenient lifestyle, and caused a series of problems: car theft every year, traffic accidents occur from time to time... There is no doubt that cars need standardized management. Now, our car management is done by humans. It's easy to imagine manual helplessness in the face of an ever-growing team of cars. Therefore, intelligent transportation will become an inevitable trend in the development of future traffic management.
Traffic intelligence realizes that you cannot be sure it will automatically identify. "As early as the 1990s, car identification has attracted widespread attention around the world, and people began to study automatic identification cards - automatic recognition of vehicle license plates Related questions. A few years later, another important status symbol of cars - car logo recognition has also become a hot topic. The general method of license plate recognition: computer image processing technology to automatically extract license plate information to determine the car license plate number. The recognition rate of the offline algorithm based on the hybrid algorithm of edge histogram and template matching has reached a very high level. The current theories of license plate and car logo patterns have matured and are moving towards integration and intelligence.
Intelligent transportation management system, vehicle identification is equivalent to the "state" of the VC++ base class, and other sub-modules in the intelligent transportation management system are inherited and developed on the basis of car identity. Therefore, we believe that car identification requires a higher level of integration, and it is best to be embedded into other systems, such as microcontrollers and CPLDs. At this stage, most of the car identification is done by computers.
In addition, due to the automatic identification and positioning of the base class, using "can only lock cars" and "can quickly determine which cars will have certain requirements at this stage." Vehicle recognition, but relies solely on a simple license plate recognition. The subject of the market is a separate license plate or vehicle identification system, the combination of the two systems is very rare. It is obviously very difficult for these single systems to achieve the purpose of truly recognizing the identity of the locked car.
Combining the requirements of intelligent traffic management systems, the status quo of today's vehicle identification, and the two development trends, the Group chose the identity of the innovative experimental plan topic of the embedded automatic vehicle identification system . It is planned to complete the embedded processing of vehicle identification and transfer the digital information to other modules in the intelligent transportation management system, but to process the vehicle number identification with the embedded computer, which will greatly improve the integration of intelligent transportation management systems and reduce costs. Different from a single recognition system, the vehicle recognition system is designed not to combine license plate recognition with vehicle recognition, and is supplemented by vehicle color recognition.
Simultaneous confirmation and output methods are used to determine and lock the car, striving to be foolproof. Which greatly facilitates use of all areas of the system.
In the field of public security traffic management, it can be applied to embedded car automatic identification systems, traffic control systems, tachometers used in embedded products, measuring overload and other transportation conveniences. You can complete a series of management; connection To the terminal computer processing system, the processed digital information is transmitted instead of image information, which significantly saves the processing time and memory space of the terminal computer, improves the response speed and processing efficiency, and effectively solves the current manpower shortage problem in the field of traffic control. .
In the park's vehicle management, the automatic identification system embeds the car's identity as it leaves the port, so it can check the park-registered vehicle with the owner connected to the resource library. Installed at the gate of the park, the automatic license plate recognition system can automatically identify vehicles entering and exiting, and then use the data in the database and the license plate data in the database to determine whether the parking lot is in the parking lot, and then process it. This will greatly improve the car safety factor of the park. Using this The cost of the system is far less than the cost of a computer processing system.
Parking lot management and embedded automatic license plate recognition system can be completed during the intelligent management process. The system is installed at the entrance of the parking lot, automatically identifies the vehicles in the parking lot, will process the data through the computer terminal, enter the information in the database to determine whether they are combined by the computer terminal, and buy (or rent) ) motor vehicle parking spaces to handle accordingly.
In short, we have reason to believe that the embedded automatic license plate recognition system we plan to complete can play a decisive role in the future intelligent traffic management system, which is worthy of research and discussion.
BR />
Appendix 2: Project Concept
The vehicle recognition system includes license plate recognition, car color and car subject recognition. The system will use embedded The system consists of three parts to complete the identification. Because of the content of our section, this idea is not very mature.
Our works and programs consist of three parts: sub-license plate recognition and body color, and car logo pattern recognition embedded in our works and programs.
: License plate recognition
1. Overall structure
The automatic license plate recognition system is mainly divided into three modules: (1) Trigger: entrance speed measurement of front-end equipment data system. (2) Part of image processing: divided into four parts: image acquisition, license plate positioning, character segmentation and character recognition. (3) The wireless transmission system sends the processed data to back-end application systems, such as traffic violation management systems, parking systems, security systems, etc.
2. Algorithm part
①Pre-end CCD camera:
Original image acquisition
CCD camera and auxiliary lighting equipment, acquisition Image quality will directly affect the performance of back-end processing and recognition. To get clearer images, you need to consider many factors that affect image quality, including: selection of cameras and frame grabbers, camera position, distance between calibration cars, speed of access units, weather, light, etc., effect Light, exposure camera.
Determine whether the vehicle enters the observation area
Use the grayscale image difference method to determine whether the monitored target area enters the first video image, and then compare the corresponding pixels of the two images The amount of gray value changes, if any.
Poor image can only be measured by monitoring objects in the field, but whether it is a transport vehicle remains to be seen. The noise generated after completion of the image is poor, pedestrians and bicycles are occupied than areas occupied by cars, the scale filter is designed to filter smaller objects and noise.
②License plate positioning and preprocessing
Left license plate positioning algorithm. The basis of license plate positioning, but also requires basic preprocessing of license plate numbers.
Tilted straightening with rivets and border removal.
I, license plate character tilt correction
In some license plates, the license plate character segmentation is difficult and the direct splitting is invalid and needs to be corrected. First, we calculate the speed of the license plate tilt and rotate the license plate based on the tilt correction.
II, license plate frame and rivet removal
Prior knowledge: standard license plate, the spacing between characters is 12mm, the spacing between characters 2 and 3 is 34mm, and the middle point is l0mm wide Small dots 2,3 character spacing 12mm.
There are usually 4 rivets on the inside of the border line of the license plate, and the first two characters, or the first six characters, are adhered to varying degrees. If the rivets are not removed, it will cause difficulty in identifying the characters 2 and 6. After the license plate image is binarized, the image is only a black and white binary file. A white pixel (gray value 255), a black pixel (gray value 0) 0, here is the black and white pattern of the license plate image being scanned line by line from the inside to the outside, when scanning the line of the license plate image, the width of the white pixel is When the edge of a license plate character is greater than a threshold (the first qualifying line), remove all lines above or below this line.
③License plate character segmentation
The algorithm segments the license plate
characters displayed in the picture.
Our limited knowledge
Do not describe these algorithms in detail
④Character recognition method... />Character
Identify the core part of car
brand recognition
.
Vehicles
All permitted characters are known
Includes 6 non-inclusive algorithms.
List
rights.
We are more interested in character recognition algorithms based on neural networks. Below, we specifically introduce two relatively simple, general algorithms, as well as character recognition algorithms based on neural networks.
I, License Plate Character Recognition by Template Matching
Chinese license plate character templates are divided into Chinese characters, English letters and numeric templates, constructed using statistical methods, and saved to the database. Template matching character templates and standardized license plate character matching to identify characters.
Second, license plate character recognition with function matching
The license plate recognition method has many personality characteristics, which can be roughly divided into structural characteristics, pixel distribution characteristics and other characteristics.
Here, we intend to focus on breakthrough neural network methods, because artificial neural network technology is suitable for large-scale parallel distributed processing capabilities, high robustness and self-learning association functions. The specific steps are as follows:
In addition, we will try to combine various algorithms to avoid the simulation and online control of nonlinear time-varying large systems. Weaknesses such as: combining genetic algorithm and artificial neural network, using the parallel computing of genetic algorithm, global search can be quickly utilized. The neural network in search can overcome the inherent shortcomings of being very slow and easily falling into local dryness.
We are still a professional basic course for sophomore students. It is not enough to understand the latest image processing algorithms. We will choose an optimal solution in actual operation, combined with our system functions. , put forward suggestions for improvement. / a>
Part 2: Car color and logo car making
①, Body color recognition
The size of the color feature depends on the image itself, Direction, angle, and other small, powerful advantages have very important applications in content-based image retrieval technology and intelligent transportation systems, as well as a large number of I-system industries (such as paper, textile, printing, etc.). For a long time, due to various reasons, a large number of color space models have been proposed, which can be mainly divided into three categories: The first category is based on the human visual system (Human Visual Variant System, H VS). This method includes Color spaces on RGB, H SI, M UNSELL color space; the second category is color spaces based on specific applications, including the adoption of YUV and YIQ in television systems, the photography industry, Kodak's printing system YCC, CMY (K) color space ;The third category is the CIE color space (CIE XYZ, CIE Laboratory, CIE LUV, etc.). The advantages and disadvantages of these color spaces, and the important roles they play in their respective fields.
RGB color space, RG?B is intended to be used by our system.
Color space is widely used in computer-related fields, such as common CRT monitors, each color value in the RGB color space? What is the combination of the values ??of the R, G, and B channels? Commonly, the corresponding channel value is passed through the photoreceptor in the image acquisition card or CCD sensor, and other similar devices, where, the channel value? Incident light and its corresponding photoreceptor sensitivity function value and expression: R =
G =
B =
where S (A), G (A), the spectrum of R(A), and B(A) are the sensitivity functions of R, G, and B of the sensor. As can be seen from the above formula, color space is a calculation in the computer, so it is device dependent, associated with the photosensitive capabilities of the specific capture device, however, since RGB values ??are easily available and can usually be used Indicates other color spaces, transformed RGB values? The standard color difference of the RGB color space for other color space values ??is defined as:
)
The subjective perception of different colors by people, in order to use the color recognition subsystem of the color difference, Color difference formula experience to better express:
The car body color recognition system we intend to design mainly includes the following four steps to complete
1. Identification area selection for body color recognition
Is it necessary to accurately identify areas? The choice of body color is confirmed. Facing the previous experiment, select the fan part near the exhaust of the car
2. Color Histogram Calculation
Calculate the color occurrence of the selected area. In real applications, due to other component values? The color space model can be represented as an RGB value as a simple calculation, the calculation of the color histogram is only for the RGB color space model.
3. Color difference calculation
The color template for calculating color difference is based on the calculation formula of chromatic aberration of each color space model.
4. Color recognition
The results of the color space model of the sample color and the standard color are based on the color recognition, that is, selecting the corresponding component calculated in the previous step Aberration, as the lowest value of the recognition result.
②, vehicle logo recognition
It is indisputable that real-time recognition of automatic license plates and vehicle bodies is crucial to an accurate identification system of motor vehicle types. The license plate positioning algorithms that have been proposed can be divided into two categories: license plate positioning algorithms based on black and white images and license plate positioning algorithms based on color images. Black and white images can be divided into many types, such as the license plate positioning algorithm based on the adaptive energy filter, the license plate positioning algorithm based on the adaptive energy filter, the license plate positioning algorithm based on the combination of binary wavelet transform and morphological processing projection. Algorithm and license plate location algorithm based on genetic algorithm.
License plate positioning algorithms all have their advantages and disadvantages, but they are the benchmark for vehicle logo positioning to a certain extent.
The positioning and identification of vehicle logos at home or abroad is a relatively new field. Largely similar car logos have inherent particularities: the target, size and lighting effects, and background are not uniform, and the shape and size of car logos are inconsistent for different car companies, making accurate positioning and recognition difficult.
The main steps of license plate positioning, which are divided into car logo recognition, are as follows:
(L): According to the texture characteristics of the license plate, the license plate area is quickly obtained based on multi-resolution analysis;
(2) Front positioning: OTSU binary image binarization algorithm, based on the front area? Higher energy is getting thicker, and then binary projection is used, combined with the license plate positioning information, to quickly position forward;
(3) Axis positioning: In the front area, position the front axle symmetrically according to the axis; < /p>
(4) Car standard rough positioning: Based on the prior knowledge of the car logo and license plate before positioning, a car logo empirical search rectangle is obtained;
(5) Car logo positioning accuracy Based on the first step (4), use the texture features of the car standard to accurately locate the main body of the vehicle. The evening consists of two steps: the vehicle logo area has the characteristics of high energy and relative concentration in the vertical direction. The energy used to enhance the vehicle logo recognition system is an important part of the motor vehicle recognition system. License plate recognition also includes two companies. Positioning and recognition of morphological filtering and adaptive car temporal positioning problems; improved template matching algorithm for accurate positioning of car logos. key technologies.
The picture shows the logo pattern of a car
In other systems, a typical target recognition system, including a structural schematic diagram of the recognition process during online and offline training. During the training process, the manually collected images of standard samples of cars are normalized, scaled and standardized before treatment, in order to obtain the car logo standard template library and template extraction. The car logo standard template library template is not only used for car logo positioning, but also can be used for feature extraction in order to identify and other characteristics, and is used in the car standard model library of cars. The positioning process requires in addition to the imported image to input the license plate position information of the vehicle. Since various automobile standards do not have a stable texture feature, which is the same size and shape as , it is very difficult to locate the car logo in the complex background of feature matching or direct template matching, so prior information must be fully utilized. , license plate positioning, vehicle symmetric rough positioning, and precise positioning based on the use of image processing technology and template matching. Rear vehicle logo positioning The car logo recognition problem is transformed into a 2D shape recognition problem, which can be achieved through template matching. However, actual image collection is often affected by light, noise, and local occlusion, shaping similar problems to traditional template matching. It is difficult to achieve satisfactory recognition using this method, so a suitable feature extraction and recognition method is usually needed to assist in car mark recognition and improve the recognition rate.
Part 3: Embedded BR />On the basis of a long history, the basic requirement of universality, the embedded system should be defined as: "a special computer system with embedded object system. "Embedded" and " "Private" and "computer system" are the three basic elements of an embedded system. The target system is embedded in the host system of the embedded system. The core of the embedded system is an embedded microprocessor, which has four major Advantages:
(1) Has a strong ability to support real-time and multi-tasking, capable of multi-tasking and shortening interrupt response time to a minimum, thus enabling internal code and real-time operations System execution time;
(2) Strong memory protection function
(3) Scalable processor architecture that can be rapidly expanded to meet high-performance embedded microprocessors.
(4) The power consumption of embedded microprocessors is very low, especially in embedded systems that rely on battery power, especially in portable wireless and mobile computing and communication equipment, the power consumption is only The era of increasingly scarce and expensive energy that can even μW milliwatts is undoubtedly tempting.
In addition, embedded real-time operating systems improve the reliability of the system. These are worthy of our making. License plate recognition system.
Usually license plate and car logo recognition algorithms take into account a large amount of calculations at the same time to meet real-time requirements. Therefore, we are prepared to use a 32-bit ARM embedded microprocessor as the core. The unit, timing control unit CPLD, is based on ARM 9 S3C 241 C embedded image acquisition and processing system, based on the embedded Linux operating system, grass, fully utilizing ARM's equipment, capabilities and low power consumption characteristics to achieve parallel data bus/ USB data interface image access, fast image processing, local storage of compressed image information and small size of IP-based data transmission allow the entire system to be simplified and made less resource intensive.
The system design includes the USB image acquisition subsystem of the entire system, the ARM processing subsystem and the network data transmission subsystem. The camera captures real-time video data and transfers it to the ARM processing board through USB. The ARM processing board has an embedded Linux operating system and fast imaging algorithm processing. Image sequence, and take appropriate measures according to the processing of the results, the network transmission subsystem can process the data and upload it to the monitoring center for further follow-up processing. The system structure is as shown in the figure.
ARM's image processing subsystem uses S3C. 2410 processor, USB image access that meets the required image processing speed, can ensure image transfer speed, expands 64M SD RAM and 64M flash memory, high-capacity RAM can save multiple images to facilitate image analysis and processing, wireless Data network management of network interfaces.
Of course, the above are just our initial ideas, these ideas are demonstrated and optimized in large-scale experiments!
Appendix 3: Schedule
1. It will take about 15 days to buy some basic supplies for the experiment.
2.
Make time to learn the necessary knowledge.
3. The project was completed in about 7 months and solved the software problem.
It took about a year to complete the hardware side and the company produced the prototype.
5.
6.