31 Lane Detection Using Image Processing for Driving Assistance
325
conditions and shadow effects. The road images were taken under normal daylight
conditions and this has a minor effect on the contrast and the intensity of the image.
The image was taken during the daylight, i.e. light directly obtained from the sun,
whereas the effect of the shadow of any object interfering with the road image contrast
is minor. The accuracy of processing the image for the standard threshold value is
high. An accurate Hough transform graph and Hough transform peak value for 17
road images were obtained.
Hardzeyeu and Klefenz [4] were also using the Hough transform methodology
but with an advanced computer vision technology in order to develop a sufficient
and robust system for driving assistance. The Hough transform has been used by the
authors as pattern recognition to enhance the locating image shape performances.
The main advantage of this method is that it is tolerant to gaps in the feature boundary
description and relatively is unaffected by the image noise. The Hough transformation
is being applied after the contrast adaptation is executed.
Chang and Lin [5] represent the vision module for lane detection in a difficult
situation such as bad weather conditions, shadow effect and fog. The canny edge
detector is applied to investigate the boundaries where the boundary image is divided
into sub-images in order to remove the noise. This sub-image is being applied with
the high contrast of colours for the road surface, and then a multi-adaptive thresholds
method is being applied for each block. The results are significantly better to solve
the different problem part of the image that has a different contrast of lane marking.
31.3 Methodology
In this section, the overall procedures of implementation, see Fig. 31.1 for details.
The algorithm used in this project was developed by using the MATLAB software.
31.3.1 Image Capture
Method 1 requires a raw image that is taken from the camera. The camera is mounted
inside the vehicle before the windshield with a 45-degree position, taking an image
of the centre of the road. The captured image shows the painted road line, including
the surrounding or the background from the vehicle. This is the main input image
that is required to be executed with the image processing software. The image is
called from its folder and is then saved in a pre-determined directory. The image
must be saved in the format of jpg and gif formats, as only this format can be read
by MATLAB. The original image is being reduced to 620 × 480 pixels to reduce the
processing time (Fig. 31.2).
325
conditions and shadow effects. The road images were taken under normal daylight
conditions and this has a minor effect on the contrast and the intensity of the image.
The image was taken during the daylight, i.e. light directly obtained from the sun,
whereas the effect of the shadow of any object interfering with the road image contrast
is minor. The accuracy of processing the image for the standard threshold value is
high. An accurate Hough transform graph and Hough transform peak value for 17
road images were obtained.
Hardzeyeu and Klefenz [4] were also using the Hough transform methodology
but with an advanced computer vision technology in order to develop a sufficient
and robust system for driving assistance. The Hough transform has been used by the
authors as pattern recognition to enhance the locating image shape performances.
The main advantage of this method is that it is tolerant to gaps in the feature boundary
description and relatively is unaffected by the image noise. The Hough transformation
is being applied after the contrast adaptation is executed.
Chang and Lin [5] represent the vision module for lane detection in a difficult
situation such as bad weather conditions, shadow effect and fog. The canny edge
detector is applied to investigate the boundaries where the boundary image is divided
into sub-images in order to remove the noise. This sub-image is being applied with
the high contrast of colours for the road surface, and then a multi-adaptive thresholds
method is being applied for each block. The results are significantly better to solve
the different problem part of the image that has a different contrast of lane marking.
31.3 Methodology
In this section, the overall procedures of implementation, see Fig. 31.1 for details.
The algorithm used in this project was developed by using the MATLAB software.
31.3.1 Image Capture
Method 1 requires a raw image that is taken from the camera. The camera is mounted
inside the vehicle before the windshield with a 45-degree position, taking an image
of the centre of the road. The captured image shows the painted road line, including
the surrounding or the background from the vehicle. This is the main input image
that is required to be executed with the image processing software. The image is
called from its folder and is then saved in a pre-determined directory. The image
must be saved in the format of jpg and gif formats, as only this format can be read
by MATLAB. The original image is being reduced to 620 × 480 pixels to reduce the
processing time (Fig. 31.2).