Vehicle localization apparatus and method

1. A vehicle localization apparatus comprising:

an image input unit configured to obtain a stereo image while a vehicle is driven;

a matching vector extracting unit configured to extract a matching vector from the stereo image while the vehicle is driven; and

a real-time localization unit configured to load a map vector of a current location based on global positioning system (GPS) information in previously constructed map data and estimate a current location of the vehicle using the extracted matching vector and the loaded map vector,

wherein the matching vector extracting unit comprises:

a region of interest (ROI) setting unit configured to divide the stereo image into a gray image and a depth image and set an ROI;

an edge image converting unit configured to convert the gray image into an edge image;

a line component detecting unit configured to extract a line component from the edge image; and

an image fusing unit configured to fuse the gray image, from which the line component is extracted, with the depth image.

2. The apparatus of claim 1 , wherein the matching vector extracting unit further comprises:

a dominant depth extracting unit configured to extract one or more dominant depth values from the fused image; and

a matching vector extractor configured to extract the matching vector for the one or more dominant depth values from the stereo image.

3. The apparatus of claim 2 , wherein the matching vector extracting unit is configured to:

extract a horizontal cumulative vector which is a matching vector for each of the one or more dominant depth values from the stereo image; and

extract a vertical cumulative vector which is a matching vector for each of the one or more dominant depth values from the stereo image.

4. The apparatus of claim 1 , wherein the real-time localization unit comprises:

a map vector loading unit configured to load the map vector of the current location based on GPS information in the map data;

a vector matching unit configured to match the extracted matching vector with the map vector; and

a vehicle localization unit configured to estimate a positioning coordinate corresponding to the extracted matching vector as a current location of the vehicle when the extracted matching vector and the map vector meet a condition.

5. The apparatus of claim 4 , wherein the vector matching unit is configured to:

extract a depth value, corresponding to a dominant depth value extracted from the stereo image while the vehicle is driven, from the map data;

extract a map vector, corresponding to a horizontal cumulative vector obtained from the stereo image while the vehicle is driven, from the map data; and

extract a map vector, corresponding to a vertical cumulative vector obtained from the stereo image while the vehicle is driven, from the map data, when a map vector corresponding to the horizontal cumulative vector meets a reference value.

6. A vehicle localization method comprising:

obtaining a stereo image while a vehicle is driven;

extracting a matching vector from the stereo image while the vehicle is driven;

loading a map vector of a current location based on GPS information in previously constructed map data;

matching the extracted matching vector with the map vector; and

estimating a positioning coordinate corresponding to the extracted matching vector as a current location of the vehicle when the extracted matching vector and the map vector meet a condition,

wherein extracting the matching vector comprises:

dividing the stereo image into a gray image and a depth image;

converting the gray image into an edge image;

extracting a line component from the edge image; and

fusing the gray image, from which the line component is extracted, with the depth image.

7. The method of claim 6 , wherein extracting the matching vector further comprises:

extracting one or more dominant depth values from the fused image; and

extracting the matching vector for the one or more depth values from the stereo image.

8. The method of claim 7 , wherein extracting the matching vector for the one or more dominant depth values comprises:

extracting a horizontal cumulative vector that is a matching vector for each of the one or more dominant depth values from the stereo image; and

extracting a vertical cumulative vector that is a matching vector for each of the one or more dominant depth values from the stereo image.

9. The method of claim 8 , wherein matching the extracting matching vector with the map data comprises:

extracting a depth value, corresponding to a dominant depth value extracted from the stereo image while the vehicle is driven, from the map data;

extracting a map vector, corresponding to the horizontal cumulative vector obtained from the stereo image while the vehicle is driven, from the map data; and

extracting a map vector, corresponding to the vertical cumulative vector obtained from the stereo image while the vehicle is driven, from the map data, when a map vector corresponding to the horizontal cumulative vector meets a reference value.

10. The method of claim 9 , wherein estimating the current location of the vehicle comprises estimating a positioning coordinate corresponding to the dominant depth value as the current location of the vehicle, when a map vector corresponding to the vertical cumulative vector meets a reference value.