Object recognition system, and obstacle recognition system and method for vehicle

• US 8,848,055 B2
• Filed: 02/22/2011
• Issued: 09/30/2014
• Est. Priority Date: 11/23/2010
• Status: Active Grant

• Current Assignee: Hyundai Mobis Co., Ltd.
• Original Assignee: Hyundai Mobis Co., Ltd.
• Inventors: Lee, Hong Seok
• Primary Examiner(s): Torrente, Richard
• Assistant Examiner(s): HUANG, FRANK F
• Application Number:
  US13/031,935
• Publication Number:
  US 20120127309A1
• Time in Patent Office: 1,316 Days
• Field of Search: 348143-160
• US Class Current: 348/148
• CPC Class Codes: G06V 10/25 Determination of region of interest [ROI] or a volume of interest [VOI]
• CPC Class Codes: G06V 10/50 by performing operations within image blocks; by using histograms, e.g. histogram of oriented gradients [HoG]; by summing image-intensity values; Projection analysis
• CPC Class Codes: G06V 20/58 Recognition of moving objects or obstacles, e.g. vehicles or pedestrians; Recognition of traffic objects, e.g. traffic signs, traffic lights or roads
• CPC Class Codes: H04N 7/18 Closed-circuit television [CCTV] systems, i.e. systems in which the video signal is not broadcast

Abstract

Provided is an object recognition system. The object recognition system recognizes an object in an ROI of a source image. The object recognition system includes an image change unit and an ROI detection unit. The image change unit receives the source image, and changes the object into an edge image which is represented as an edge line. The ROI detection unit divides the edge image into a plurality of regions, compares a total sum of edge component values of an edge line included in each of the regions and a predetermined threshold value by regions, and detects a region, in which the total sum of edge component values is greater than the threshold value, as the ROI from among the plurality of regions.

Claims

1. An obstacle recognition system for a vehicle, comprising:

an image photographing unit comprised in the vehicle, and configured to generate a single source image which is obtained by photographing a front view or a rear view while the vehicle is driving;

an edge image generation unit configured to receive the source image, and generate an edge image which represents an object comprised in the source image as an edge line;

a region of interest (ROI) detection unit configured to divide the edge image into a predetermined number of regions according to a speed value of the vehicle, calculate a total sum of edge component values of the edge line for each region, compare the total sum of edge component values of each of the regions and a predetermined threshold value for each of the regions, and detect a region from the number of regions as an ROI that provides information of the object in front of the vehicle on the basis of the compared result for each of the regions; and

an image analysis unit configured to scan the detected ROI by block units having a certain size to analyze whether an obstacle exists in the detected ROI,

wherein the ROI detection unit comprises;

an image divider configured to receive the speed value of the vehicle from a vehicle system, and divide an entire region of the edge image into a predetermined number of regions according to the received speed value of the vehicle;

an edge value collector configured to calculate a total sum of edge component values of an edge line which is comprised in each of the regions, and collect the calculated total sum of edge component values of each of the regions; and

a comparator configured to compare the total sum of edge component values of the edge line of each of the regions and the predetermined threshold value for each region, and detect a region, in which the total sum of edge component values is greater than the threshold value, as the ROI from among the plurality of regions,

wherein the image analysis unit is configured to scan only the detected ROI without scanning any other region of the plurality of regions in order to reduce a processing time of the scanning,

wherein when an entire size of the edge image is i*j, i being the number of pixels in a horizontal direction, and j being the number of pixels in a vertical direction, a size of each of the divided regions is i*(j/n), n being a natural number from two to six,

wherein as a speed of the vehicle increases, the n increases, in order to decrease a detection error occurring in an operation of detecting the ROI when the entire region of the edge image is divided into equal portions irrespective of a vehicle speed.

2. The obstacle recognition system of claim 1, wherein the image divider is configured to receive the speed value of the vehicle from the vehicle system in a controller area network (CAN) communication scheme.

3. The obstacle recognition system of claim 1, wherein the image divider is configured to divide the edge image into n equal portions in a horizontal direction in a predetermined size according to the speed value of the vehicle, n being a natural number equal to or more than 2.

4. The obstacle recognition system of claim 1, wherein the edge image generation unit, the ROI detection unit and the image analysis unit are configured to be modularized in a chip type.

5. An obstacle recognition method for a vehicle which scans a region of interest (ROI) detected from a single source image by block units having a certain size to recognize an obstacle, the obstacle recognition method comprising:

generating an edge image that represents an object comprised in the source image as an edge line;

dividing the edge image into a predetermined number of regions according to a speed value of the vehicle; and

calculating a total sum of edge component values of an edge line that exists in each of the divide regions, comparing the total sum of the edge component values and a predetermined threshold value, and detecting a region from the number of regions as a ROI that provides information of the object in front of the vehicle on the basis of the compared result for each of the regions,

wherein as the speed value of the vehicle increases, the predetermined number increases in order to reduce a processing time of the scanning and decrease a detection error occurring in an operation of detecting the ROI when the entire region of the edge image is divided into equal portions irrespective of a vehicle speed.

6. The obstacle recognition system of claim 5, wherein the edge image is divided in a horizontal direction.