As car industry is growing, pedestrian traffic accident is increasing. To reduce the traffic accidents and the severity of the pedestrian
injuries, there have been many researches such as road infrastructure improvement, car structure improvement, and advanced driver
assistance system (ADAS) based on sensor network. There are many sensors to detect the pedestrian, for example, IR sensor, LADAR,
proximity sensor, temperature sensor, image sensor, and so on. Among them, the image sensor is the most intuitive and includes much more
information compare to the other sensors. Furthermore, because vehicles equipped with image recording device is generalized, it is
essential to study the pedestrian detection using an image sensor.
In this thesis, we propose a pedestrian detection algorithm using Gabor filter bank. In general, a Gabor filter is a linear filter used for
edge detection, because frequency and orientation representations of Gabor filters are similar to those of human visual system.
In order to extract the features of the pedestrian, we use various image processing algorithms and data structure algorithms. First, color
image segmentation is performed to consider the information of the RGB color space. Second, histogram equalization is performed to
enhance the brightness of the input images. Third, convolution is performed between a Gabor filter bank and the enhanced images.
Fourth, statistical values are calculated by using the integral image(summed area table) method. The calculated statistcal values
such as mean, variance, skewness are used for the feature matrix of the pedestrian area. To evaluate the proposed algorithm, the INRIA pedestrian database
and support vector machine(SVM) are used. The SVM classifies with the non-pedestrian or pedestrian using the extracted feature matrix.
We evaluate the proposed algorithm by using the INRIA test database and road images captured with a camera installed in a
vehicle. The experimental results show that the proposed algorithm is more accurate compare to the histogram of oriented gradient(HOG)
pedestrian detector which is called the methodology of pedestrian detection algorithm at present. Optimization of the proposed algorithm and ensuring the real-time
capability is our further works.