Method of vehicle segmentation and counting for nighttime video frames

• US 8,019,157 B2
• Filed: 10/09/2008
• Issued: 09/13/2011
• Est. Priority Date: 06/23/2008
• Status: Active Grant

• Current Assignee: Huper Laboratories Co., Ltd.
• Original Assignee: Huper Laboratories Co., Ltd.
• Inventors: Chang, Chao-Ming, Chen, Jun-Liang, Chen, Chao-Ho
• Primary Examiner(s): Tran; Phuoc
• Application Number:
  US12/248,054
• Publication Number:
  US 20090316957A1
• Time in Patent Office: 1,069 Days
• Field of Search: 382/103, 382/104, 382/107, 382/162, 382/167, 382/173, 382/260, 382/286, 382/291, 348/148, 348/149
• US Class Current: 382/173
• CPC Class Codes: G06T 2207/30236 Traffic on road, railway or crossing
• CPC Class Codes: G06T 2207/30242 Counting objects in image
• CPC Class Codes: G06T 7/215 Motion-based segmentation
• CPC Class Codes: G06V 20/584 of vehicle lights or traffic lights

Abstract

Vehicle segmentation and counting method based on the property of color variation and headlight information combining change detection in nighttime traffic environment (one-way road with coming direction) is provided. The goal is to reduce the effect of ground-illumination that decreases the accuracy of vehicle segmentation. Besides, the amount of traffic flow is calculated and it can be used in other post-applications, such as traffic-flow reporting or controlling.

Claims

1. A method of vehicle segmentation and counting for night video frames, comprising:

generating a background frame according to the video frames;

converting the video frames from RGB data to HSI data;

converting the background frame from RGB data to HSI data;

inputting the HSI data of the video frames and the HSI data of the background frame to a Gaussian filter so as to generate Gaussian data of the video frames and Gaussian data of the background frame;

updating the Gaussian data of the background frame;

determining changed regions and unchanged region of the video frames to generate a frame difference mask and a background subtraction mask according to the Gaussian data of the video frames and the Gaussian data of the background frame;

detecting an object region of the video frames to generate a static region mask and a moving region mask according to the frame difference mask and the background subtraction mask;

combining the static region mask and the moving region mask to generate an initial object mask;

detecting characteristic information of the moving object to calculate the amount of the moving objects;

excluding an interference region of the moving object and compensating the object region of the video frames to generate a final object mask; and

obtaining the moving object from the video frames according to the final object mask.

2. The method of claim 1, wherein the video frames are nighttime video frames.

3. The method of claim 1, wherein the moving object is a vehicle, and the characteristic information is the headlight information of the vehicle.

4. The method of claim 1, wherein the interference region comprises a shadow region and a ground-illumination region.

5. The method of claim 1, wherein excluding the interference region of the moving object and compensating the object region of the video frames to generate the final object mask is excluding the interference region of the moving object and compensating the object region of the video frames to generate the final object mask according to the characteristic information.