Moving Object Detection Based on Optical Flow Estimation and a Gaussian Mixture Model for Advanced Driver Assistance Systems

Abstract

Most approaches for moving object detection (MOD) based on computer vision are limited to stationary camera environments. In advanced driver assistance systems (ADAS), however, ego-motion is added to image frames owing to the use of a moving camera. This results in mixed motion in the image frames and makes it difficult to classify target objects and background. In this paper, we propose an efficient MOD algorithm that can cope with moving camera environments. In addition, we present a hardware design and implementation results for the real-time processing of the proposed algorithm. The proposed moving object detector was designed using hardware description language (HDL) and its real-time performance was evaluated using an FPGA based test system. Experimental results demonstrate that our design achieves better detection performance than existing MOD systems. The proposed moving object detector was implemented with 13.2K logic slices, 104 DSP48s, and 163 BRAM and can support real-time processing of 30 fps at an operating frequency of 200 MHz.

Keywords: ADAS; FPGA; background subtraction; moving object detection; optical flow estimation.

Figure 1

Overall scheme of the proposed moving object detection (MOD) algorithm.

Figure 2

Compensation for the integer parts of the ego-motion. The shaded region denotes empty space generated by the shift operation: (a) ${\mu }_{n,t-1}$ memory; (b) $w_{n,t-1}$ memory; (c) ${\sigma }_{n,t-1}^2$ memory.

Figure 3

Block diagram of the proposed moving object detector.

Figure 4

Hardware structure: (a) optical flow estimator; (b) convolution calculator.

Figure 5

Block diagram of the resolution process unit.

Figure 6

Hardware structure of the camera motion estimator.

Figure 7

Block diagram of the background detector.

Figure 8

Block diagram of the object detector.

Figure 9

FPGA test platform: (a) test environment; (b) Xilinx Virtex-5 FPGA based evaluation board; (c) 640 × 480 resolution camera.

Figure 10

MOD performance of the proposed moving object detector.