Vehicle detecting method, nighttime vehicle detecting method based on dynamic light intensity and system thereof

• US 10,878,259 B2
• Filed: 11/23/2018
• Issued: 12/29/2020
• Est. Priority Date: 10/17/2018
• Status: Active Grant

• Current Assignee: Automotive Research & Testing Center
• Original Assignee: Automotive Research & Testing Center
• Inventors: Liao, Po-Hsiang, Lin, Hung-Pang, Chang, Yu-Lung, Hsu, Li-You
• Primary Examiner(s): Flores, Leon
• Application Number:
  US16/198,896
• Publication Number:
  US 20200125869A1
• Time in Patent Office: 767 Days
• Field of Search: None
• CPC Class Codes: G06T 2207/10024 Color image
• CPC Class Codes: G06T 2207/20021 Dividing image into blocks, subimages or windows
• CPC Class Codes: G06T 2207/20076 Probabilistic image processing
• CPC Class Codes: G06T 2207/30252 Vehicle exterior; Vicinity of vehicle
• CPC Class Codes: G06T 7/246 using feature-based methods, e.g. the tracking of corners or segments
• CPC Class Codes: G06T 7/269 using gradient-based methods
• CPC Class Codes: G06T 7/77 using statistical methods
• CPC Class Codes: G06T 7/90 Determination of colour characteristics
• CPC Class Codes: G06V 10/507 Summing image-intensity values; Histogram projection analysis
• CPC Class Codes: G06V 10/60 relating to illumination properties, e.g. using a reflectance or lighting model
• CPC Class Codes: G06V 20/584 of vehicle lights or traffic lights

Abstract

A nighttime vehicle detecting method is for capturing an image by a camera and driving a computing unit to compute the image and then detect a highlight point of the image. The nighttime vehicle detecting method is for driving the computing unit to perform a communicating region labeling algorithm to label a plurality of highlight pixels connected to each other as a communicating region value, and then performing an area filtering algorithm to analyze an area of the highlight pixels connected to each other and judge whether the highlight pixels connected to each other are a vehicle lamp or not according to a size of the area. The nighttime vehicle detecting method is for driving the computing unit to perform an optical flow algorithm to obtain a speed of the vehicle lamp, and then filtering the vehicle lamp moved at the speed smaller than a predetermined speed.

Claims

1. A nighttime vehicle detecting method based on dynamic light intensity, comprising:

providing a highlight detecting step, wherein the highlight detecting step is for capturing an image by a camera and driving a computing unit to compute the image and then detect a highlight point of the image, the highlight point comprises a plurality of highlight pixels, and the camera and the computing unit are disposed on a driving vehicle;

providing a vehicle lamp judging step, wherein the vehicle lamp judging step is for driving the computing unit to perform a communicating region labeling algorithm to label the highlight pixels connected to each other as a communicating region value, and then performing an area filtering algorithm to analyze an area of the highlight pixels connected to each other and judge whether the highlight pixels connected to each other are a vehicle lamp or not according to a size of the area;

providing an optical flow filtering step, wherein the optical flow filtering step is for driving the computing unit to perform an optical flow algorithm to obtain a speed of the vehicle lamp, and then filtering the vehicle lamp moved at the speed smaller than a predetermined speed; and

providing a distance estimating step, wherein the distance estimating step is for driving the computing unit to perform a coordinate conversion algorithm to estimate a distance between the vehicle lamp and the camera;

wherein the highlight detecting step comprises;

providing a histogram equalization step, wherein the histogram equalization step is for counting a number of occurrences of each of a plurality of gray scale values of the image, and changing the gray scale values to generate a plurality of equalized gray scale values according to the number of occurrences.

2. The nighttime vehicle detecting method of claim 1, wherein the highlight detecting step further comprises:

providing a brightness feature extracting step, wherein the brightness feature extracting step is for setting a gray scale brightness threshold value and extracting each of the highlight pixels whose brightness value is higher than the gray level brightness threshold value.

3. The nighttime vehicle detecting method of claim 1, wherein the histogram equalization step comprises:

a cumulative distribution function for changing the gray scale values, wherein the cumulative distribution function comprises the gray scale values, the number of occurrences, a gray level and the equalized gray scale values, the gray scale values are represented as i, the number of occurrences is represented as n_i, the gray level is represented as L, the equalized gray scale values are represented as T_i, and the cumulative distribution function is described as follows;

4. The nighttime vehicle detecting method of claim 1, wherein,

in the highlight detecting step, the computing unit judges whether the highlight pixels are the vehicle lamp or not according to a color of the highlight pixels;

when the color of the highlight pixels is red, the computing unit determines that the highlight pixels are the vehicle lamp which is a taillight; and

when the color of the highlight pixels is white, the computing unit determines that the highlight pixels are the vehicle lamp which is a headlight.

5. The nighttime vehicle detecting method of claim 1, wherein,

in the vehicle lamp judging step, the highlight pixels are divided into a plurality of highlight pixel groups, the highlight pixels of each of the highlight pixel groups are connected to each other, the computing unit is configured to perform the communicating region labeling algorithm to label the highlight pixel groups as a plurality of communicating region values, respectively, and then the computing unit is configured to perform the area filtering algorithm to analyze the area of each of the highlight pixel groups and judge whether each of the highlight pixel groups is the vehicle lamp or not according to the size of the area.

6. The nighttime vehicle detecting method of claim 1, wherein,

in the vehicle lamp judging step, the computing unit is configured to perform a geometric proportion analyzing algorithm to analyze a geometric proportion of the highlight pixels labeled as the communicating region value, and then judge whether the highlight pixels labeled as the communicating region value are the vehicle lamp or not according to the geometric proportion.

7. A nighttime vehicle detecting system based on dynamic light intensity, comprising:

a camera disposed on a driving vehicle and configured to capture an image; and

a computing unit disposed on the driving vehicle and comprising;

a highlight detecting module signally connected to the camera, wherein the highlight detecting module is configured to compute the image and then detect a highlight point of the image, and the highlight point comprises a plurality of highlight pixels;

a vehicle lamp judging module signally connected to the highlight detecting module, wherein the vehicle lamp judging module is configured to perform a communicating region labeling algorithm to label the highlight pixels connected to each other as a communicating region value, and then perform an area filtering algorithm to analyze an area of the highlight pixels connected to each other and judge whether the highlight pixels connected to each other are a vehicle lamp or not according to a size of the area;

an optical flow filtering module signally connected to the vehicle lamp judging module, wherein the optical flow filtering module is configured to perform an optical flow algorithm to obtain a speed of the vehicle lamp, and then filter the vehicle lamp moved at the speed smaller than a predetermined speed; and

a distance estimating module signally connected to the optical flow filtering module, wherein the distance estimating module is configured to perform a coordinate conversion algorithm to estimate a distance between the vehicle lamp and the camera;

wherein the highlight detecting module is configured to count a number of occurrences of each of a plurality of gray scale values of the image, and change the gray scale values to generate a plurality of equalized gray scale values according to the number of occurrences.

8. The nighttime vehicle detecting system of claim 7, wherein,

the highlight detecting module is configured to set a gray scale brightness threshold value and extract each of the highlight pixels whose brightness value is higher than the gray level brightness threshold value.

9. The nighttime vehicle detecting system of claim 7, wherein,

the highlight detecting module is configured to perform a cumulative distribution function for changing the gray scale values, the cumulative distribution function comprises the gray scale values, the number of occurrences, a gray level and the equalized gray scale values, the gray scale values are represented as i, the number of occurrences is represented as n the gray level is represented as L, the equalized gray scale values are represented as T_i, and the cumulative distribution function is described as follows:

10. The nighttime vehicle detecting system of claim 7, wherein,

the computing unit judges whether the highlight pixels are the vehicle lamp or not according to a color of the highlight pixels;

when the color of the highlight pixels is red, the computing unit determines that the highlight pixels are the vehicle lamp which is a taillight; and

when the color of the highlight pixels is white, the computing unit determines that the highlight pixels are the vehicle lamp which is a headlight.

11. The nighttime vehicle detecting system of claim 7, wherein,

the highlight pixels are divided into a plurality of highlight pixel groups, the highlight pixels of each of the highlight pixel groups are connected to each other, the vehicle lamp judging module is configured to perform the communicating region labeling algorithm to label the highlight pixel groups as a plurality of communicating region values, respectively, and then the vehicle lamp judging module is configured to perform the area filtering algorithm to analyze the area of each of the highlight pixel groups and judge whether each of the highlight pixel groups is the vehicle lamp or not according to the size of the area.

12. The nighttime vehicle detecting system of claim 7, wherein,

the vehicle lamp judging module is configured to perform a geometric proportion analyzing algorithm to analyze a geometric proportion of the highlight pixels labeled as the communicating region value, and then judge whether the highlight pixels labeled as the communicating region value are the vehicle lamp or not according to the geometric proportion.

13. A vehicle detecting method for detecting a front vehicle of an image, the vehicle detecting method comprising:

providing an image analyzing step, wherein the image analyzing step is for capturing an image by a camera and driving a computing unit to analyze a sky brightness value of a sky region of the image and compare a predetermined brightness value with the sky brightness value to judge whether the image is in a daytime or a nighttime, and the camera and the computing unit are disposed on a driving vehicle; and

providing a vehicle detecting step, wherein the vehicle detecting step is for driving the computing unit to perform a daytime vehicle detecting step or a nighttime vehicle detecting step, when the image is in the daytime, the daytime vehicle detecting step is performed, and when the image is in the nighttime, the nighttime vehicle detecting step is performed;

wherein the daytime vehicle detecting step is for detecting a distance between the front vehicle and the camera according to a forward collision warning algorithm;

wherein the nighttime vehicle detecting step comprises;

providing a highlight detecting step, wherein the highlight detecting step is for driving the computing unit to compute the image and then detect a highlight point of the image, and the highlight point comprises a plurality of highlight pixels;

providing a vehicle lamp judging step, wherein the vehicle lamp judging step is for driving the computing unit to perform a communicating region labeling algorithm to label the highlight pixels connected to each other as a communicating region value, and then performing an area filtering algorithm to analyze an area of the highlight pixels connected to each other and judge whether the highlight pixels connected to each other are a vehicle lamp or not according to a size of the area;

providing an optical flow filtering step, wherein the optical flow filtering step is for driving the computing unit to perform an optical flow algorithm to obtain a speed of the vehicle lamp, and then filtering the vehicle lamp moved at the speed smaller than a predetermined speed; and

providing a distance estimating step, wherein the distance estimating step is for driving the computing unit to perform a coordinate conversion algorithm to estimate a distance between the vehicle lamp and the camera;

wherein the highlight detecting step comprises;

providing a histogram equalization step, wherein the histogram equalization step is for counting a number of occurrences of each of a plurality of gray scale values of the image, and changing the gray scale values to generate a plurality of equalized gray scale values according to the number of occurrences.

14. The vehicle detecting method of claim 13, wherein,

in the image analyzing step, the sky region is located above a predetermined skyline;

when the sky brightness value is greater than or equal to the predetermined brightness value, the computing unit is configured to determine that the image is in the daytime; and

when the sky brightness value is smaller than the predetermined brightness value, the computing unit is configured to determine that the image is in the nighttime.

15. The vehicle detecting method of claim 13, wherein the highlight detecting step further comprises:

providing a brightness feature extracting step, wherein the brightness feature extracting step is for setting a gray scale brightness threshold value and extracting each of the highlight pixels whose brightness value is higher than the gray level brightness threshold value.

16. The vehicle detecting method of claim 13, wherein the histogram equalization step comprises:

a cumulative distribution function for changing the gray scale values, wherein the cumulative distribution function comprises the gray scale values, the number of occurrences, a gray level and the equalized gray scale values, the gray scale values are represented as i, the number of occurrences is represented as n_i, the gray level is represented as L, the equalized gray scale values are represented as T_i, and the cumulative distribution function is described as follows;

17. The vehicle detecting method of claim 13, wherein,

in the highlight detecting step, the computing unit judges whether the highlight pixels are the vehicle lamp or not according to a color of the highlight pixels;

when the color of the highlight pixels is red, the computing unit determines that the highlight pixels are the vehicle lamp which is a taillight; and

when the color of the highlight pixels is white, the computing unit determines that the highlight pixels are the vehicle lamp which is a headlight.

18. The vehicle detecting method of claim 13, wherein,

in the vehicle lamp judging step, the highlight pixels are divided into a plurality of highlight pixel groups, the highlight pixels of each of the highlight pixel groups are connected to each other, the computing unit is configured to perform the communicating region labeling algorithm to label the highlight pixel groups as a plurality of communicating region values, respectively, and then the computing unit is configured to perform the area filtering algorithm to analyze the area of each of the highlight pixel groups and judge whether each of the highlight pixel groups is the vehicle lamp or not according to the size of the area.

19. The vehicle detecting method of claim 13, wherein,

in the vehicle lamp judging step, the computing unit is configured to perform a geometric proportion analyzing algorithm to analyze a geometric proportion of the highlight pixels labeled as the communicating region value, and then judge whether the highlight pixels labeled as the communicating region value are the vehicle lamp or not according to the geometric proportion.