TW201527156A - Lane recognition method and electronic device using the same - Google Patents

Abstract

A lane recognition method and an electronic device using the same are provided. The lane recognition method includes following steps. At least one lane image in front of a car is obtained by an image capturing unit. A lane detecting procedure is executed based on the lane image. If a current driving lane can not be obtained through the lane detecting procedure, a reference license plate in the lane image is recognized. Lane sharing information transmitted by another car is received through a communication unit, and the lane sharing information and the reference license plate are compared with each other so as to obtain the current driving lane.

Description

Lane identification method and electronic device using the same

The present invention relates to a lane recognition method, and more particularly to a lane recognition method based on collaborative workshop communication and an electronic device using the same.

With the advancement of technology, transportation has become one of the important inventions in modern development. However, due to the rapid development of the city, the connection of roads has become more complicated. Therefore, the emergence of navigation systems, such as the Global Position System (GPS) navigation system and the Traffic Message Channel (TMC), brings great convenience to human beings. At present, the common car navigation device not only provides guidance services such as roads and scenic spots, but also enables the driver to understand the current driving situation, and can also locate and track the driver's position, direction and speed.

However, most of the current vehicle navigation only uses the GPS and road level map information, so it is impossible to tell the driver to switch to that lane based on the current driving lane, which makes the driver easy to be unfamiliar or complicated. of The intersection missed the opportunity to switch lanes. For example, if the driver does not immediately find out that the current driving lane is not a lane that can be turned, it is often easy to miss the destination.

If the vehicle's lane position is simply identified by the image, it will be limited by environmental factors and accurate lane recognition cannot be performed. For example, when the vehicle is in a traffic jam or a shelter obstructing the lane line, it will not be possible to identify the current vehicle passing through the image. In addition, although high-precision positioning systems, such as DGPS, can reduce positioning errors and obtain current driving lanes, high-precision positioning systems are expensive to manufacture and susceptible to weather. Therefore, how to provide a perfect and low-cost lane recognition positioning system is one of the topics of interest to those skilled in the art.

In view of this, the present invention provides a lane recognition method and an electronic device using the same, which can achieve the purpose of recognizing the current driving lane by using a low-cost device.

The invention provides a lane recognition method suitable for an electronic device disposed on a vehicle. The vehicle is equipped with an image capturing unit and a communication unit, and the lane recognition method includes the following steps. Acquiring at least one lane image in front of the vehicle through the image capturing unit. The lane detection procedure is performed by the lane image. If it is not possible to obtain the current driving lane of the vehicle through the lane detection program, the reference license plate in the lane image is recognized. The lane sharing information transmitted by another vehicle is received through the communication unit, and information is shared with the reference license plate and the lane to obtain the current driving lane.

In an embodiment of the invention, the lane image described above includes at least history The image and the current image, and the lane recognition method further comprises: analyzing a change in the slope of the at least one lane line in the historical image and the current image, thereby determining whether the vehicle changes the current driving lane. If the vehicle changes the current driving lane, the current driving lane of the vehicle is updated.

In an embodiment of the present invention, the lane recognition method further includes: broadcasting the license plate of the vehicle and the current driving lane through the communication unit, so that the other vehicle receives the license plate and the current driving lane.

In an embodiment of the invention, before the step of executing the lane detection program by the lane image, the method further comprises: positioning by using a vehicle positioning unit to obtain a vehicle position of the vehicle. According to the vehicle location and the map information, the number of lanes of the current road of the vehicle is obtained.

In an embodiment of the invention, the step of executing the lane detection program by the lane image comprises: setting at least one region of interest corresponding to the lane image. Converting the region of interest corresponding to the lane image into at least one grayscale image. Performing edge detection and line detection on grayscale images to obtain a plurality of lane lines. The number of lanes of the current road of the vehicle and the current driving lane are calculated based on the lane line.

In an embodiment of the invention, the step of performing edge detection and line detection on the grayscale image to obtain the lane line includes: performing edge detection and line detection on the grayscale image to obtain a plurality of A candidate line. A plurality of second candidate straight lines are filtered from the first candidate straight line according to the straight line equation of the first candidate straight line. Calculating a plurality of intersections of the second candidate line and the reference horizontal line, and determining whether the second candidate lines are lane lines according to the distance between the intersection points.

In an embodiment of the invention, if the above-mentioned lane detection process cannot be passed The step of acquiring the current driving lane of the vehicle and identifying the reference license plate in the lane image comprises: performing image pre-processing on the lane image, and locating the license plate area of the reference license plate from the lane image. Character segmentation is performed for the license plate area to divide a plurality of character images. These character images are identified to obtain a plurality of license plate characters of the reference license plate.

In an embodiment of the invention, the character images are identified by the above The step of taking the characters of the reference license plate includes comparing the character images and the sample characters to obtain a plurality of preliminary decision characters represented by the respective character images. It is judged whether these preliminary decision characters belong to a specific category, and these preliminary decision characters belonging to a specific category are re-identified by the character feature recognition program, thereby acquiring a plurality of license plate characters of the reference license plate.

In an embodiment of the invention, the receiving the other one through the communication unit The lanes shared by the vehicle share information and share the information with the reference license plate and the lane, so that the step of obtaining the current driving lane includes: analyzing the current driving of the vehicle according to the relative position between the other vehicle and the vehicle and the lane sharing information. Lane.

From another point of view, the present invention provides an electronic device, the electronic device Placed on a vehicle. The electronic device is coupled to the image capturing unit, and the electronic device includes a communication unit, a memory, and a processor. The memory is stored with multiple instructions. The processor is coupled to the communication unit and the memory for executing the instructions to perform the following steps. Acquiring at least one lane image in front of the vehicle through the image capturing unit. The lane detection procedure is performed by the lane image. If it is not possible to obtain the current driving lane of the vehicle through the lane detection program, the reference license plate in the lane image is recognized. Communication unit The lane sharing information transmitted by another vehicle is received, and information is shared with the reference license plate and the lane to obtain the current driving lane.

Based on the above, in an embodiment of the present invention, a lane image in front of the vehicle is acquired by the image capturing unit, and image processing and analysis are performed on the lane image. When the necessary information for detecting the lane in the lane image is incomplete or obscured by the obstacle, the license plate of the preceding vehicle can be identified by the lane image. Accordingly, by sharing the information of the license plate information of the preceding vehicle with the lanes transmitted by other vehicles through the workshop communication, the current location of the vehicle can be known from the lane sharing information of the preceding vehicle based on the comparison result to further obtain the current state of the vehicle. Driving lane.

The above described features and advantages of the invention will be apparent from the following description.

10‧‧‧ Lane Identification System

100‧‧‧Electronic devices

110‧‧‧ memory

111‧‧‧ Lane Detection Module

112‧‧‧ License Plate Identification Module

113‧‧‧ Lane Tracking Module

120‧‧‧ processor

200‧‧‧Image capture unit

300‧‧‧Vehicle locating unit

400‧‧‧Communication unit

R1‧‧‧ Road

LN1~LN3‧‧" lane

C_1~C_3‧‧‧ Vehicles

Img1~Img7‧‧‧ images

M1~M4, T1~T2‧‧‧ lane lines

Z1~Z2‧‧‧ interest area

H1‧‧‧projection chart

L1~L4‧‧‧Sampling line segment

S1‧‧‧ similar character group

I1‧‧‧Judgement Mechanism

S301~S306‧‧‧ steps of the lane recognition method according to an embodiment of the present invention

S401~S409‧‧‧ steps of the lane recognition method according to another embodiment of the present invention

FIG. 1 is a schematic diagram of a lane recognition system according to an embodiment of the invention.

2 is a schematic diagram of a lane recognition method according to an embodiment of the invention.

FIG. 3 is a flow chart of a lane recognition method according to an embodiment of the invention.

4 is a flow chart of a lane recognition method according to another embodiment of the present invention.

FIG. 5 is a schematic diagram of setting an area of interest according to another embodiment of the invention.

FIG. 6A is a schematic diagram of an overlay grayscale image according to another embodiment of the invention.

FIG. 6B is a schematic diagram of an edge detection image according to another embodiment of the invention.

FIG. 6C is a schematic diagram of eliminating image noise according to another embodiment of the invention.

FIG. 7 is a schematic diagram of character cutting according to another embodiment of the present invention.

FIG. 8 is a schematic diagram of character recognition according to another embodiment of the present invention.

FIG. 9 is a schematic diagram of setting an area of interest according to another embodiment of the invention.

FIG. 10 is a schematic diagram of a tracking lane line according to another embodiment of the invention.

1 is a diagram of a lane recognition system according to an embodiment of the invention. intention. Referring to FIG. 1 , the lane recognition system 10 includes an electronic device 100 , an image capturing unit 200 , a vehicle positioning unit 300 , and a communication unit 400 . The lane recognition system 10 is applied to a vehicle, and the image capturing unit 200 is disposed in front of the vehicle, such as above the windshield of the vehicle, to capture the image of the lane in front of the vehicle. Generally, the image capturing unit 200 can be disposed in or outside the vehicle, and the present invention is not limited thereto. The image capturing unit 200 has, for example, a charge coupled device (Charge Coupled) Device, CCD) or Complementary Metal-Oxide Semiconductor (CMOS) component image sensor for capturing lane images in front of the vehicle. For example, the image capturing unit 200 may be a driving recorder having a video recording device or a camera having a photography function. The number and configuration of the image capturing unit 200 can be changed according to actual needs, and is not limited herein.

The vehicle positioning unit 300 is, for example, a Global Position System (GPS), which can locate the current position of the vehicle and calculate driving information such as the traveling direction and speed. The communication unit 400 is configured to transmit and receive data in a wireless network and has corresponding hardware settings in response to the type of wireless network. The present invention is not limited to the implementation of the communication unit 400, and any wireless communication device that can receive or transmit data through wireless transmission is within the scope of the present invention. Further, the communication unit 400 can communicate with other vehicles based on a specific communication protocol, for example, based on the vehicle communication protocol under the IEEE 1609 standard, but the present invention is not limited thereto.

The electronic device 100 includes at least a processor 120 and a memory 110. The electronic device 100 can be implemented as a computer, a car computer, or other vehicle electronic device. For example, the vehicle electronic device may be a car navigation device, but the invention is not limited to such implementations. The processor 120 can be, for example, a central processing unit (CPU), a microprocessor (Microprocessor), an application specific integrated circuit (ASIC), or a programmable logic device (PLD). Or other hardware devices with computing power. The memory 110 is, for example, a random access memory (random An access memory, a flash memory or other memory for storing data and a plurality of modules, and the processor 120 is coupled to the memory 110 and configured to execute the modules. The modules include a lane detection module 111, a license plate recognition module 112, and a lane tracking module 113. These modules are, for example, computer programs that can be loaded into the processor 120 to perform the function of recognizing lanes.

2 is a diagram showing a lane recognition method according to an embodiment of the invention. intention. Referring to FIG. 2, it is assumed that the road R1 has three lanes, which are the first lane LN1, the second lane L2, and the third lane LN3, respectively. The vehicle C_1, the vehicle C_2, and the vehicle C_3 travel on the road R1, and the vehicle C_1, the vehicle C_2, and the vehicle C_3 are located on the second lane LN2 of the road R1. As shown in FIG. 2, the vehicle C_2 is immediately behind the vehicle C_1, and the vehicle C_3 is immediately behind the vehicle C_2. In the exemplary embodiment shown in FIG. 2, the vehicle C_2 and the vehicle C_3 are too close to the preceding vehicle, so that the current lane information cannot be obtained by analyzing the lane image. In contrast, since there is no blocking vehicle in front of the vehicle C_1, the vehicle C_1 can obtain the current lane information by analyzing the lane image. Once the vehicle C_1 knows the current lane position, the vehicle C_1 will broadcast the lane sharing information, and the lane sharing information of the vehicle C_1 includes at least the current driving lane (the second lane LN2) and the license plate number of the vehicle C_1.

In the present invention, when the vehicle cannot obtain the correct one by its own analysis In the lane information, since the two vehicles adjacent to each other are located in the same lane, the current driving lane of the preceding vehicle can be used as the current driving lane of the rear vehicle. Taking Figure 2 as an example, although the vehicle C_2 cannot analyze the current driving lane by itself, the vehicle C_2 can pass through. The wireless transmission receives lane sharing information transmitted by other vehicles. Furthermore, the vehicle C_2 can recognize the license plate number of the preceding vehicle (vehicle C_1) from the lane image to select the lane sharing information belonging to the vehicle C_1 from the plurality of lane sharing information by the identification result. Based on this, the vehicle C_2 can know the current driving lane of the vehicle C_1 from the lane sharing information of the vehicle C_1, and further use the current driving lane of the vehicle C_1 as its current driving lane. Similarly, when the vehicle C_2 knows the current lane position, the vehicle C_2 will broadcast the lane sharing information, so that the vehicle C_3 can also know the current lane position by the license plate recognition and the lane sharing information of the vehicle C_2.

In order to further illustrate the present invention, the following detailed description of the detailed steps of the electronic device 100 performing the lane recognition method will be described. FIG. 3 is a flow chart showing a lane recognition method according to an embodiment of the invention. Referring to FIG. 1 and FIG. 3 simultaneously, in the embodiment, the lane recognition method can be performed by using the electronic device 100 in FIG. 1, for example. The steps of the lane recognition method of the present embodiment will be described below in conjunction with various elements in the lane recognition system 10.

First, in step S301, the lane detection module 111 acquires at least one lane image in front of the vehicle through the image capturing unit 200. In step S302, the lane detection module 111 performs a lane detection program by using the lane image. In other words, in the present embodiment, the lane detection module 111 can perform lane detection by using the lane image captured by the image capturing unit 200. It should be noted that if the lane detection is to be performed by analyzing the lane image, the information carried in the lane image must be complete, otherwise the lane detection module 111 will not easily analyze the lane information through the lane image. Generally speaking, when the vehicle is in front of the open space and can completely obtain the lane line information, the lane detection The module 111 can capture the lane image captured by the image capturing unit 200. However, if the vehicle is in a traffic jam, because the distance between each car in the lane is very close, the lane detection module 111 may not be able to obtain the current lane information because the lane line is obscured by the adjacent vehicle or the preceding vehicle, or Because the image is too complicated, it produces incorrect analysis information. In other words, lane detection by analyzing lane images will be limited by many environmental factors.

Therefore, in step S303, the lane detection module 111 determines whether it is transparent. The lane detection program acquires the current driving lane of the vehicle. In detail, the lane detection module 111 will determine whether to acquire the lane information of the vehicle and check whether the analyzed lane information is correct. For example, in one embodiment, the lane detection module 111 is positioned by the vehicle positioning unit 300 to obtain a vehicle location on the map. Then, the lane detection module 111 acquires the number of lanes of the currently traveling road according to the vehicle location and the map information. In other words, based on the positioning system and the map information, the lane detection module 111 can know the number of lanes of the currently traveling road. Therefore, the lane detection module 111 determines whether the number of lanes in the map information is the same as the number of lanes detected by the lane image. If yes, the representative of the analysis information is correct, the lane detection module 111 can obtain the current driving lane of the vehicle through the image processing-based lane detection program. If not, indicating that the analysis information is incorrect, the lane detection module 111 determines that the current driving lane of the vehicle cannot be acquired through the image processing-based lane detection program.

If the determination in step S303 is YES, the representative lane detection module 111 can acquire the current driving lane of the vehicle through the image processing-based lane detection program. Thereafter, in step S306, the lane detection module 111 broadcasts the included vehicle through the communication unit 400. The license plate shares information with the lane of the current driving lane, so that another vehicle receives the license plate and the current driving lane. That is to say, when the lane detection module 111 has analyzed the correct lane information by using the lane image, the lane detection module 111 uses the communication unit 400 to broadcast the location of the lane of the vehicle and its own license plate information. In this way, other vehicles can receive the lane sharing information broadcasted by the communication unit 400, so that other vehicles can know the lane position by the lane sharing information broadcasted by the communication unit 400.

On the other hand, if the current driving lane of the vehicle cannot be acquired through the lane detecting program, the license plate recognition module 112 recognizes the reference license plate in the lane image in step S304. In other words, in the case of the lane congestion, although the electronic device 100 cannot obtain accurate lane information through the image, the license plate recognition module 112 can still detect and recognize the license plate of the preceding vehicle in the lane image. Next, in step S305, the license plate recognition module 112 receives the lane sharing information transmitted by another vehicle through the communication unit 400, and compares the reference license plate with the lane to share information, thereby acquiring the current driving lane. In detail, the license plate recognition module 112 may receive the lane sharing information transmitted by many vehicles through the communication unit 400, but the license plate recognition module 112 needs the lane information of the previous adjacent vehicle. Therefore, since the license plate recognition module 112 has learned the license plate number of the preceding vehicle through the lane image, the license plate recognition module 112 can filter the lane sharing information transmitted by the preceding vehicle by comparing the identification result with the license plate information in the lane sharing information. . If the license plate information in the lane sharing information matches the identification result, the license plate recognition module 112 can learn the current driving lane from the filtered lane sharing information.

It is worth mentioning that the implementation manner of the present invention is not limited to the above description, and The contents of the above embodiments are changed as appropriate for actual needs. For example, after the electronic device 100 acquires the lane information by using the lane detection program or the license plate recognition, the electronic device 100 can further perform a lane tracking program by using the lane image to confirm whether the vehicle switches the current driving lane. Hereinafter, another embodiment of the present invention will be described in detail to describe the steps of the lane recognition method.

FIG. 4 is a diagram of a lane recognition method according to another embodiment of the invention. flow chart. Referring to FIG. 1 and FIG. 4 simultaneously, in the embodiment, the lane recognition method can be performed by using the electronic device 100 in FIG. 1, for example. The steps of the lane recognition method of the present embodiment will be described below in conjunction with various elements in the lane recognition system 10.

First, in step S401, the lane detection module 111 transmits image through Unit 200 acquires at least one lane image in front of the vehicle. In step S402, the lane detection module 111 performs a lane detection program by using the lane image. In step S403, the lane detection module 111 determines whether the current driving lane of the vehicle is acquired through the lane detection program.

In the lane detection program of this embodiment, the lane detection module 111 first At least one Region of Interest (ROI) corresponding to the lane image is set, and the region of interest corresponding to the lane image is converted into at least one grayscale image. In detail, since the image information required for detecting the lane line is the lane line on the ground, the lane detecting module 111 selects the region of interest in the lane image as a specific region covering at least all the lane lines. For example, FIG. 5 is a schematic diagram of a selected region of interest according to another embodiment of the invention. Referring to FIG. 5, the lane detection module 111 Searching and setting the region of interest Z1 below a baseline (e.g., horizon) on the lane image Img1. As shown in FIG. 5, the region of interest Z1 is a specific region covering all of the lane lines M1 to M4. Based on the selection of the interest area Z1, the operation speed can be improved and the unnecessary noise interference can be reduced.

Then, the lane detection module 111 converts the selected region of interest into a grayscale image, and the lane detection module 111 performs edge detection and line detection on the converted grayscale image to obtain a plurality of lanes. line. In detail, based on the obvious characteristics of the lane line edge feature, the lane detection module 111 can use edge detection to find the edge of the lane line. It is worth mentioning that in an embodiment, the lane line in the region of interest may be too short or discontinuous due to the presence of a dotted line on the lane line. Therefore, after the lane detection module 111 converts the extracted region of interest into a grayscale image, the lane detection module 111 can superimpose a plurality of grayscale images adjacent to the sampling time to make the lane line become a layer in the superimposed image. A more continuous and distinct line.

For example, FIG. 6A is a schematic diagram of superimposed grayscale images according to another embodiment of the invention. FIG. 6B is a schematic diagram of an edge detection image according to another embodiment of the invention. In the example shown in FIG. 6A, the image Img2 is a superimposed image generated by the lane detecting module 111 superimposing four consecutive lane images. In the example shown in FIG. 6B, the image Img3 is an edge image generated by the lane detection module 111 for edge detection of the image Img2.

In addition, in an embodiment, in order to eliminate some obvious noise, the lane detection module 111 can use the erosion (Erosion) in Morphology to eliminate noise. For example, FIG. 6C illustrates a second embodiment of the present invention. A schematic diagram that eliminates image noise. In the example shown in FIG. 6C, the image Img4 is an image generated by the lane detection module 111 performing morphological erosion on the edge image Img3.

Furthermore, the lane detection module 111 utilizes edges generated by edge detection. Image for line detection. In one embodiment, the lane detection module 111 detects a line in the edge image by using Hough line detection, for example. However, in other embodiments, the lane detection module 111 may use other algorithms to perform line detection to further find a straight edge in the region of interest that may be the edge of the lane line, which is not limited by the present invention. In this embodiment, the lane detection module 111 performs edge detection and line detection on the grayscale image of the region of interest to obtain a plurality of first candidate lines.

Based on this, the lane detection module 111 can be judged by a specific judgment mechanism. Whether these first candidate lines are lane lines. In the embodiment in which the first candidate line is detected by using the Hough line detection, the lane detection module 111 may filter the plurality of second lines from the first candidate lines according to the line equations of the first candidate lines. Candidate line. Next, the lane detection module 111 calculates a plurality of intersections of the second candidate line and the reference horizontal line, and determines whether the second candidate line is a lane line according to the distance between the intersection points.

In detail, since the Hoff line detection uses two coordinate points on the image to represent The straight line segments searched by the table, so the lane detecting module 111 first calculates a straight line equation representing each line segment by using two punctuation points of each line segment. Thereafter, the lane detection module 111 determines whether the line is a lane line by a straight line equation of each line segment. Due to the car The lane lines are not horizontal straight lines, so the lane detection module 111 can remove the first candidate line that is a horizontal line segment according to the slope of the first candidate lines. In addition, if there is a first candidate line having the same straight line equation, the lane detecting module 111 may retain only one of them. Based on this, the lane detection module 111 can filter out the second candidate line from among the plurality of first candidate lines.

After that, the lane detection module 111 calculates all the second candidate lines and fixed The intersection of the height reference horizontal lines, and the intersections are arranged in order. Since the distance between the lane lines (the lane width) falls within a certain reasonable interval, the lane detection module 111 can calculate the distance between the intersections to detect the lane line from the second candidate line. Based on the detection and determination of the lane line, the lane detection module 111 will find the straight line segment most likely to be the lane line from the region of interest of the lane image.

Based on the lane line, the lane detection module 111 can calculate the target of the vehicle. The number of lanes on the road ahead and the current lane. Further, the total number of lane lines minus one is the number of lanes on the current road. For example, if the lane detection module 111 finds 4 lane lines, it represents that the current road is a three lane. Furthermore, when the image capturing unit 200 is placed in front of the vehicle, the slope of the lane line located on the left side of the vehicle must be a positive value. Conversely, the slope of the lane line on the right side of the vehicle must be negative. That is to say, by analyzing the slope of the lane line, the lane detecting module 111 can know that there are several lane lines on the left and right sides of the vehicle. In this way, the lane detecting module 111 can further know which lane the vehicle is located by the number of the left lane line and the right lane line, thereby acquiring the current driving lane.

Please refer to FIG. 4 again. If the determination in step S403 is no, in step S404. The license plate recognition module 112 recognizes the reference license plate in the lane image. In this embodiment, the license plate recognition process includes image pre-processing, license plate location, character cutting, and character recognition. In one embodiment, the license plate recognition module 112 performs image pre-processing on the lane image and locates the license plate area of the reference license plate from the lane image. The license plate recognition module 112 performs character segmentation for the license plate area to divide a plurality of character images. The license plate recognition module 112 then recognizes the character images to obtain a plurality of license plate characters of the reference license plate.

In detail, the license plate recognition module 112 first performs image pre-processing to strengthen The characteristics of the license plate in the lane image make it easier to locate the license plate position and size after the license plate is positioned. For example, in order to make the edges in the lane image more visible and easier to detect, the license plate recognition module 112 can enhance the contrast of the image by using histogram equalization. Histogram equalization is to increase the grayscale of all pixels in the image as evenly as possible from 0 to 255 to improve the contrast of the image.

In the license plate location, the license plate recognition module 112 utilizes the grayscale in the license plate image. The characteristics of the drastic changes to find out the area that may be the license plate, and then use the projection cutting method to achieve the purpose of initial license plate positioning. For example, the license plate recognition module 112 uses the edge detection side to find all the edge features in the lane image, and uses the morphological expansion and erosion to combine all the edge parts into one piece, which is convenient for calculating the license plate in the image. position. Then, the license plate recognition module 112 respectively projects white points (pixel points representing dramatic changes in gray scale) in the edge image generated by the image pre-processing onto the horizontal axis and the vertical axis, and the license plate recognition module 112 further separates The cumulative number of white spots falling on the horizontal and vertical axes is counted. Based on a relatively large number of white dots representing the license plate in the image In a general location, the license plate recognition module 112 can position the license plate. In an embodiment, the license plate recognition module 112 can further extend the image of the license plate taken out, and use the method of projection cutting to finely adjust the boundary of the license plate image.

After the license plate position is located, the license plate recognition module 112 performs the functions of character cutting and character recognition. That is to say, the license plate recognition module 112 cuts out each character image from the license plate image, and then recognizes each character after cutting each character image. It should be noted that any character segmentation technique can be used to implement the character slice of the present invention, which is not limited by the present invention.

For example, FIG. 7 is a schematic diagram of character cutting according to another embodiment of the present invention. In the character cutting method exemplified in FIG. 7, the license plate recognition module 112 performs cutting according to the space between the characters. As shown in FIG. 7, the license plate recognition module 112 first projects the license plate image Img5 onto the horizontal axis to generate a statistical map H1 of the grayscale value projection amount. It can be known that since there is a space between the characters, the amount of grayscale value corresponding to the blank area between the characters must be the highest. Based on this, as shown in FIG. 7, the peaks P1 to P5 in the graph H1 can be regarded as the position of one character. That is to say, as shown by the image Img6 in FIG. 7, the license plate recognition module 112 can initially cut out the respective characters according to the left and right edges of the license plate and the peaks P1 to P5. In addition, the license plate recognition module 112 can further obtain the width B1 of the character by the cut position C1 on the left side of the first character and the cut position C2 on the left side of the second character, and further cut each of the characters according to the width B1 of the character. Character.

After the character cutting, the license plate recognition module 112 can perform the step of character recognition. In an embodiment, the license plate recognition module 112 uses, for example, a pattern ratio. For (Template matching) to compare the characters, the characters to be recognized and all the sample images are aligned one by one. The purpose of identification is achieved by taking the sample image closest to the character to be recognized as the recognition result. For example, the grayscale values of two images of the same size are subtracted, and the smaller the difference of the grayscale values after subtraction, the closer the two images are. Among them, the character database for identification can be established by collecting sample images of all letters and numbers first. It should be particularly noted that although the above embodiment is described by taking a sample alignment as an example, those skilled in the art can implement the steps of recognizing license plate characters by any character recognition technology.

In addition, after comparing the preliminary decision characters using the pattern, these characters are compared due to, for example, '0', '6', '8', '9', 'C', 'D', and 'S'. Similar, so as long as the character image is slightly offset or tilted, it may cause a recognition error. Therefore, in an embodiment, after the license plate recognition module 112 performs character recognition by sample comparison, the license plate recognition module 112 performs feature comparison again on the similar characters to increase the correct rate of character recognition. .

That is to say, the license plate recognition module 112 first compares the cut character image and the sample character to obtain the preliminary determination character represented by each character image. Next, the license plate recognition module 112 determines whether the preliminary decision characters belong to a particular category consisting of similar characters. For example, if the preliminary decision character belongs to similar characters such as '0', '6', '8', '9', 'C', 'D', and 'S', the license plate recognition module 112 Then it is necessary to identify again through specific features. In other words, the license plate recognition module 112 can re-identify a plurality of preliminary determination characters belonging to a specific category through the character feature recognition program, thereby acquiring the vehicle of the reference vehicle license plate of the preceding vehicle. Card character.

FIG. 8 is a schematic diagram of character recognition according to another embodiment of the invention. In the character recognition method exemplified in FIG. 8, in order to distinguish each character in the similar character group S1, the license plate recognition module 112 respectively extracts the sampling line segments L1 to L4 on the character image, and respectively determines the character characters. Whether the sampling line segment L1 to the sampling line segment L4 is a white line segment. Based on the structural difference of each character, the license plate recognition module 112 can distinguish the characters in the similar character group S1 according to whether the sampling line segment L1 to the sampling line segment L4 are white line segments. For example, for the number '6', since the upper right side has a gap, the sampling line segment L3 on the number '6' is a white line segment, and the remaining sampling line segments L1, L2, and L4 are not white line segments. And so on, the judgment mechanism established based on the above rules is as shown in the judgment mechanism table I1 in FIG.

In other words, based on the judgment mechanism table I1 shown in FIG. 8, the license plate recognition module 112 can distinguish the similar characters in the similar character group S1 by the sampling line segments L1 to L4 on the character image. It is worth mentioning that, as shown in the judgment mechanism table I1, the number '0' and the letter 'D' cannot be distinguished by the judgment mechanism of the judgment mechanism table I1. Therefore, the license plate recognition module 112 can perform discrimination again by using the characteristic difference between the upper left corner and the lower left corner of the number '0' and the letter 'D'. Thereby, the license plate recognition module 112 can accurately identify all the license plate characters in the license plate image to know the license plate number of the preceding vehicle.

Referring to FIG. 4 again, after the license plate recognition module 112 recognizes the license plate number of the preceding vehicle, in step S405, the license plate recognition module 112 receives the lane sharing information transmitted by another vehicle through the communication unit 400, and compares the reference license plate. Share with lane Information to get the current driving lane. Different from the embodiment shown in FIG. 3, in the present embodiment, once the lane detection module 111 or the license plate recognition module 112 obtains the lane position, the lane tracking module 113 performs lane tracking. The lane image includes a historical image and a current image. The historical image is different from the sampling time of the current image, and the current image can be regarded as the currently acquired one lane image. Then, after the license plate recognition module 112 acquires the current driving lane, in step S406, the lane tracking module 113 analyzes the slope change of the at least one lane line in the historical image and the current image, thereby determining whether the vehicle changes the current driving lane. If the vehicle changes the current driving lane, the lane tracking module 113 updates the current driving lane of the vehicle.

In detail, in an embodiment, the lane tracking module 113 can be based on the calendar. The lane line position already known in the history image is used to find a new lane line position in the current image, and it is determined whether the vehicle switches lanes based on the change in the slope. In the present embodiment, the lane tracking module 113 first selects the historical image and the region of interest of the current image as the region for tracking the lane line. Since the vehicle usually maintains a safe distance from the preceding vehicle, the left and right side lane lines near the front of the vehicle are less likely to be obscured. Therefore, FIG. 9 is a schematic diagram of setting an interest area according to an embodiment of the invention. Referring to FIG. 9, the lane tracking module 113 can set, for example, the region Z2 at the lowermost edge of the lane image Img7 as the region of interest for detection and analysis. As shown in FIG. 9, the zone Z2 includes a left lane line T1 and a right lane line T2.

In the absence of historical images, the lane tracking module 113 can take The midpoint of the lane line and the straight line equation representing the lane line is calculated by the least square regression line. After that, it was detected by historical images. After the detected lane line position, the new lane line position can be found in the current image based on the position detected by the historical image. In this way, the area of the image used for calculation is greatly reduced and the operation speed is increased. In other words, the lane tracking module 113 detects the location of the lane line in the tracking area associated with the previous information based on the previous information of at least one lane line in the historical image, thereby acquiring the current information of the lane line.

FIG. 10 is a schematic diagram of a tracking lane line according to another embodiment of the invention. Assume that the left lane line T1 in FIG. 9 is taken as an example. As shown in FIG. 10, the current image is first searched based on the left lane line found by the historical image. The dotted quadrilateral Q1 represents the historical image. The left lane line and the solid line quadrilateral Q2 represent the left lane line of the current image. Since the center line Lpre of the left lane line has been found in the historical image, the lane tracking module 113 uses the center line to averagely extract 16 reference lines respectively from the line segments Ly1 to Ly16 from the top to the bottom. Then, in the line segments Ly1~Ly16, find the left edge point (such as the point P _Left ) and the right edge point (such as the point P _Right ) of the left lane line in the current image to the left and right, and then calculate the left edge point and the right edge point. The midpoint (such as point P _mid ).

It is worth mentioning that since the lane line may be a dotted line, the line segments that do not find the left and right edges (such as the line segment Ly1 and the line segment Ly2) are excluded, and then all the left edge points and all the right edge edges are obtained from the line segment Ly3 to the line segment Ly16. The midpoint of the point. Thereafter, the lane tracking module 113 calculates a linear equation Lcur representing the position of the new left lane line using the least square regression line and the midpoints (eg, the midpoint P _mid on the line segment Ly16). By analogy, the lane tracking module 113 continuously repeats the same search action based on the new lane line position to achieve the purpose of tracking the lane line.

When the lane tracking module 113 obtains the positions of the left and right lane lines, The lane tracking module 113 determines whether the vehicle changes lanes according to the change in the slope of the lane line. For example, if the vehicle wants to switch to the left lane, the slope of the left lane line of the current lane will gradually increase from a positive value. Then, during the transition to the left lane, the slope of this left lane line is converted to a negative value and then gradually becomes larger. Similarly, when the vehicle wants to switch to the right lane, the slope of the right lane line of the current driving lane will gradually decrease from a negative value. Then, during the transition to the right lane, the slope of the right lane line becomes a positive value and then gradually becomes smaller. It can be seen that the lane tracking module 113 can detect whether the vehicle changes the lane and its changed lane position by using the change of the slope of the left and right lane lines. If it is determined that the vehicle switches lanes, the lane tracking module 113 analyzes the current driving lane after the switching according to the slope change of the lane line, and updates the current driving lane accordingly.

Next, in step S407, the license plate tracking module 113 transmits through the communication unit 400. The broadcast includes the license plate sharing information with the updated current driving lane, so that the other vehicle receives the license plate and the updated current driving lane. In other words, during the continuous tracking of the lane image through the continuously sampled lane image, the license plate tracking module 113 can continuously broadcast the updated lane information to avoid the electronic device 100 being inconsistent with the reality due to the lane change. The lane information is shared.

On the other hand, if the determination in step S403 is YES, the representative lane detection module 111 The current driving lane can be obtained directly through the lane detection program. Similarly, in step S408, the lane tracking module 113 analyzes at least one lane line in the historical image and the current image. The slope of the change is based on which it is determined whether the vehicle is changing the current driving lane. If the vehicle changes the current driving lane, the lane tracking module 113 updates the current driving lane of the vehicle. In step S409, the license plate tracking module 113 broadcasts the lane sharing information including the license plate and the updated current driving lane through the communication unit 400, so that the other vehicle receives the license plate and the updated current driving lane. The content of steps S408 to S409 is the same as or similar to that of steps S406 to S407, and details are not described herein again.

It is worth mentioning that, in addition to using the lane information of the vehicle in front to obtain the current driving lane, the electronic device 100 can also use the lane sharing information sent by the vehicle in the left front or the right front to obtain the current driving lane. That is to say, the electronic device 100 can analyze the current driving lane of the vehicle according to the relative position between the two vehicles and the lane sharing information. For example, when the license plate of the previous adjacent vehicle cannot be successfully recognized due to external factors such as staining or reflection, the electronic device 100 can recognize the license plate of another vehicle on the right front side, and utilize the right front side. The lanes sent by another vehicle share information to smoothly locate the lane location where the vehicle is located.

In summary, when the lane image is used to detect the lane, the necessary information is not complete. When the whole object is obscured by the obstacle, the present invention can recognize the license plate of the preceding vehicle by the lane image, and accordingly, the current location of the vehicle is known from the lane sharing information of the preceding vehicle to further acquire the current driving lane of the vehicle. In this way, the accuracy and convenience of the navigation system can be improved by the accurate lane recognition system. Moreover, compared with the high-precision positioning system, the hardware equipment required by the present invention is cheaper, so that the low-cost device can be used to identify the driving lane.

Although the present invention has been disclosed in the above embodiments, it is not intended to limit the present invention, and any one of ordinary skill in the art can make some changes and refinements without departing from the spirit and scope of the present invention. The scope of the invention is defined by the scope of the appended claims.

S301~S306‧‧‧ steps of the lane recognition method according to an embodiment of the present invention

Claims (18)

A lane recognition method is applicable to an electronic device disposed on a vehicle, wherein the vehicle is configured with an image capturing unit and a communication unit, and the lane recognition method includes: acquiring at least one of the front of the vehicle through the image capturing unit a lane image; performing a lane detection procedure by the lane image; if a lane of the vehicle is not available through the lane detection program, identifying a reference license plate in the lane image; and receiving through the communication unit One lane of information transmitted by another vehicle shares information and shares information with the reference license plate to the lane to obtain the current driving lane. The lane recognition method of claim 1, wherein the lane image includes at least one historical image and a current image, and the lane recognition method further comprises: analyzing the historical image and at least one lane line in the current image. a slope change according to which it is determined whether the vehicle changes the current driving lane; and if the vehicle changes the current driving lane, the current driving lane of the vehicle is updated. The lane recognition method of claim 1, wherein the lane recognition method further comprises: broadcasting, by the communication unit, a license plate including the vehicle and the lane sharing information of the current driving lane, so that another vehicle receives the The license plate is with the current driving lane. For example, the lane identification method described in claim 1 is Before the step of performing the lane detection procedure, the lane image further includes: positioning by a vehicle positioning unit to obtain a vehicle position of the vehicle; and acquiring a current status of the vehicle according to the vehicle location and a map information The number of lanes on the road. The lane recognition method of claim 1, wherein the step of executing the lane detection program by the lane image comprises: setting at least one region of interest corresponding to the lane image, and corresponding to the lane image Converting the region of interest into at least one grayscale image; performing an edge detection and line detection on the grayscale image to obtain a plurality of lane lines; and calculating a current driving path of the vehicle according to the lane line The number of lanes and the current driving lane. The lane recognition method of claim 5, wherein the step of acquiring the lane line by performing the edge detection and the line detection on the grayscale image comprises: performing the edge detection on the grayscale image And determining, by the line detection, a plurality of first candidate lines; filtering a plurality of second candidate lines from the first candidate line according to the line equation of the first candidate line; and calculating the second candidate line And a plurality of intersections with a reference horizontal line, determining whether the second candidate straight line is the lane line according to a distance between the intersection points. The lane recognition method of claim 1, wherein if the current driving lane of the vehicle cannot be acquired through the lane detection program, the step of recognizing the reference license plate in the lane image comprises: Performing an image pre-processing on the image, and positioning a license plate area of the reference license plate from the lane image; performing character segmentation on the license plate area to divide a plurality of character images; and identifying the character image To obtain a plurality of license plate characters of the reference license plate. The lane recognition method of claim 7, wherein the step of recognizing the character image to obtain the license plate character of the reference license plate comprises: comparing the character image and the sample character Obtaining a plurality of preliminary decision characters represented by each of the character images; and determining whether the preliminary decision characters belong to a specific category, and re-identifying the belonging to the specific category by using a character feature recognition program The preliminary decision character is obtained to obtain the license plate character of the reference license plate. The lane recognition method of claim 1, wherein the lane sharing information transmitted by the other vehicle is received through the communication unit, and the information is shared with the reference license plate and the lane, thereby acquiring the current driving lane. The step of analyzing includes analyzing the current driving lane of the vehicle according to the relative position between the other vehicle and the vehicle and the lane sharing information. An electronic device is disposed on a vehicle and coupled to an image capturing unit and a communication unit. The electronic device includes: a memory having a plurality of modules stored therein; and a processor coupled to the communication unit and the memory to access and execute the module recorded in the storage unit, the module comprising: a lane The detection module obtains at least one lane image in front of the vehicle through the image capturing unit, and performs a lane detection program by using the lane image; and a license plate recognition module, if the lane detection program cannot pass through the lane detection program Obtaining a current driving lane of the vehicle, the license plate recognition module identifying a reference license plate in the lane image, wherein the license plate recognition module receives a lane sharing information transmitted by another vehicle through the communication unit, and compares The reference license plate shares information with the lane to obtain the current driving lane. The electronic device of claim 10, wherein the lane image includes at least one historical image and a current image, and the module further comprises: a lane tracking module, analyzing the historical image and the current image A slope change of at least one lane line is determined to determine whether the vehicle changes the current driving lane. If the vehicle changes the current driving lane, the lane tracking module updates the current driving lane of the vehicle. The electronic device of claim 11, wherein the lane tracking module broadcasts, by the communication unit, a license plate including the license plate of the vehicle and the lane of the current driving lane, so that another vehicle receives the license plate and the vehicle Currently driving lanes. The electronic device of claim 10, wherein the lane detection module is positioned by a vehicle positioning unit to obtain a vehicle position of the vehicle, and obtain the vehicle according to the vehicle position and a map information. The number of one lane of a currently traveling road. The electronic device of claim 10, wherein the lane detection module sets at least one region of interest corresponding to the lane image, and converts the region of interest corresponding to the lane image into at least one grayscale image. And performing an edge detection and a line detection on the grayscale image to obtain a plurality of lane lines, wherein the lane detection module calculates a number of lanes of a current traveling road of the vehicle according to the lane line And the current driving lane. The electronic device of claim 14, wherein the lane detection module performs the edge detection and the line detection on the grayscale image to obtain a plurality of first candidate lines, wherein, according to the a straight line equation of the first candidate straight line, the lane detecting module filters a plurality of second candidate straight lines from the first candidate straight line, and calculates a plurality of intersection points of the second candidate straight line and a reference horizontal line, and a basis The distance between the intersection points determines whether the second candidate line is the lane line. The electronic device of claim 10, wherein the license plate recognition module performs an image pre-processing on the lane image, and locates a license plate area of the reference license plate from the lane image, wherein the license plate region The recognition module performs character segmentation on the license plate area to divide a plurality of character images. The license plate recognition module recognizes the character image to obtain a plurality of license plate characters of the reference license plate. The electronic device of claim 16, wherein the license plate recognition module compares the character image and the sample character to obtain a plurality of preliminary determination words represented by each of the character images. And the license plate recognition module determines whether the preliminary determination character belongs to a specific category, and re-identifies the preliminary determination character belonging to the specific category through a character recognition program, thereby acquiring the reference license plate The license plate character. The electronic device of claim 10, wherein the license plate recognition module analyzes the current driving lane of the vehicle according to the relative position between the other vehicle and the vehicle and the lane sharing information.