17-03-2014, 05:30 PM
Abstract—A method for threshold selection to detect edge
width of objects in the image was presented. Three order spline
wavelet was choosed to complete multiscale wavelet edge
detection and wavelet edge module values were obtained. After
properties of these data was analyzed, a method for threshold
selection was proposed using half of maximum of wavelet edge
module values. Edge width of objects in the image was given a
defination and can be calculated by area and girth of the
transition edge using the threshold selected by the method
proposed. Experiment results show that the algorithm is effective
and performs better.
Index Terms—Image Processing, Threshold Selection, Edge
Width Detection, Wavelet Transform
I. INTRODUCTION
Any image-forming system has properties of the low-pass
filter and makes objects in the image smoothed. So edges of
objects in the image are of width. There may be ideal brickwall
edge in the continuous image, but if it is sampled
according to Shannon’s theorems edge width in the discrete
image is at least one pixel. The edge in the image acquisited by
image-forming system from the ideal brick-wall edge in reality
becomes slope gradient.
In certain conditions edge width detection of objects in the
image is necessary. For example, when focus of the camera is
fixed, the objects in the image are obscured and their edge
widths are different with the distance between the objects and
the camera. So if edge width of the objects in the image is
detected, we can infer the distance between the objects and the
camera and adjust focus of the camera to make the objects in
the image become sharply focused. This technology can be
used in the automatic focused system. Edge width detection
can also be used in automatic fuzzy segmentation for infrared
vehicle target image.
Threshold selection method is critical in edge width
detection of objects in the image and determines directly
accuracy of result of edge width detection. Common threshold
selection methods, for example otsu method, hard threshold
method, soft threshold method, adaptive threshold method etc.,
are used mainly in some areas such as image segmentation,
binarizing image and so on[1,2]. They are not suitable for edge
width detection of objects in the image. We proposed a method
for threshold selection based on wavelet transform.
width of objects in the image was presented. Three order spline
wavelet was choosed to complete multiscale wavelet edge
detection and wavelet edge module values were obtained. After
properties of these data was analyzed, a method for threshold
selection was proposed using half of maximum of wavelet edge
module values. Edge width of objects in the image was given a
defination and can be calculated by area and girth of the
transition edge using the threshold selected by the method
proposed. Experiment results show that the algorithm is effective
and performs better.
Index Terms—Image Processing, Threshold Selection, Edge
Width Detection, Wavelet Transform
I. INTRODUCTION
Any image-forming system has properties of the low-pass
filter and makes objects in the image smoothed. So edges of
objects in the image are of width. There may be ideal brickwall
edge in the continuous image, but if it is sampled
according to Shannon’s theorems edge width in the discrete
image is at least one pixel. The edge in the image acquisited by
image-forming system from the ideal brick-wall edge in reality
becomes slope gradient.
In certain conditions edge width detection of objects in the
image is necessary. For example, when focus of the camera is
fixed, the objects in the image are obscured and their edge
widths are different with the distance between the objects and
the camera. So if edge width of the objects in the image is
detected, we can infer the distance between the objects and the
camera and adjust focus of the camera to make the objects in
the image become sharply focused. This technology can be
used in the automatic focused system. Edge width detection
can also be used in automatic fuzzy segmentation for infrared
vehicle target image.
Threshold selection method is critical in edge width
detection of objects in the image and determines directly
accuracy of result of edge width detection. Common threshold
selection methods, for example otsu method, hard threshold
method, soft threshold method, adaptive threshold method etc.,
are used mainly in some areas such as image segmentation,
binarizing image and so on[1,2]. They are not suitable for edge
width detection of objects in the image. We proposed a method
for threshold selection based on wavelet transform.