Edge detection with adaptive threshold
Systems and methods are provided for detecting edges in an image. In an example method, the image is smoothed using a filter. A gradient magnitude and angle are determined for each pixel of the image. A non-maximum suppression is performed on the resulting image data. A double threshold with an upper and a lower threshold are applied to the resulting image data to determine the edges in the image. The upper and the lower thresholds are calculated automatically using a distribution of intensity values of the image. In example systems, an image is acquired by a camera, or other image acquisition units. An image processing unit is configured to detect the edges in the image using an adaptive threshold edge detection method.
1. A method for detecting edges in an image, the method comprising:
smoothing the image using a filter;
determining a gradient magnitude and angle for each pixel of the image;
performing a non-maximum suppression; and
applying a double threshold with an upper and a lower threshold for determining the edges in the image, where the upper and the lower threshold are calculated automatically using a distribution of intensity values of the image determined by using a histogram of the image divided into a predetermined number of bins;
where:
a normalized density function is determined for each bin resulting in a normalized probability density function, and
the upper and the lower threshold are determined by using the following equations:
T 1 =max(0,(μ−σ)/ a )
T 2 =min( A, (μ+σ)/ b ) where:
μ is the mean value of the normalized probability density function,
σ is the variance of the probability density function,
a is a parameter, between 5 and 7, preferably 6,
b is a parameter between 2 and 4, preferably 3, and
A is the maximum intensity value.
2. A computer program product having computer program code stored on a computer-readable non-transitory storage medium, the computer program code adapted to be executed to implement a method comprising the steps of:
smoothing the image using a filter;
determining a gradient magnitude and angle for each pixel of the image;
performing a non-maximum suppression; and
applying a double threshold with an upper and a lower threshold for determining the edges in the image, where the upper and the lower threshold are calculated automatically using a distribution of intensity values of the image determined by using a histogram of the image divided into a predetermined number of bins;
where:
a normalized density function is determined for each bin resulting in a normalized probability density function, and
the upper and the lower threshold are determined by using the following equations:
T 1 =max(0,(μ−σ)/ a )
T 2 =min( A, (μ+σ)/ b ) where:
μ is the mean value of the normalized probability density function,
σ is the variance of the probability density function,
a is a parameter between 5 and 7, preferably 6,
b is a parameter between 2 and 4, preferably 3, and
A is the maximum intensity value.
3. A computer-readable non-transitory storage medium having a computer program stored thereon, the computer program code adapted to be executed to implement a method comprising the steps of:
smoothing the image using a filter;
determining a gradient magnitude and angle for each pixel of the image;
performing a non-maximum suppression; and
applying a double threshold with an upper and a lower threshold for determining the edges in the image, where the upper and the lower threshold are calculated automatically using a distribution of intensity values of the image determined by using a histogram of the image divided into a predetermined number of bins;
where:
a normalized density function is determined for each bin resulting in a normalized probability density function, and
the upper and the lower threshold are determined by using the following equations:
T 1 =max(0,(μ−σ)/ a )
T 2 =min( A, (μ+σ)/ b ) where:
μ is the mean value of the normalized probability density function,
σ is the variance of the probability density function,
a is a parameter between 5 and 7, preferably 6,
b is a parameter between 2 and 4, preferably 3, and
A is the maximum intensity value.
4. An imaging system comprising:
an image acquisition unit configured to acquire an image;
an image processing unit configured to detect edges in the image by smoothing the image, determining a gradient magnitude and angle for each pixel of the image, performing a non-maximum suppression and applying a double threshold to determine the edges, where the image processing unit automatically calculates the upper and the lower threshold using a distribution of intensity values of the image determined by using a histogram of the image divided into a predetermined number of bins;
where:
a normalized density function is determined for each bin resulting in a normalized probability density function, and
the upper and the lower threshold are determined by using the following equations:
T 1 =max(0,(μ−σ)/ a )
T 2 =min( A, (μ+σ)/ b ) where:
μ is the mean value of the normalized probability density function,
σ is the variance of the probability density function,
a is a parameter between 5 and 7, preferably 6,
b is a parameter between 2 and 4, preferably 3, and
A is the maximum intensity value.
5. The method of claim 1 where the mean value and the variance are determined as a first or second derivative of the normalized probability density function, respectively.
6. The method of claim 1 where the mean value, μ, is determined by:
μ
=
∑
i
=
1
N
f
norm
(
i
)
·
x
(
i
)
,
where:
N is a number of bins,
f norm (i) is a normalized density function for each bin i, and
x(i) is a center intensity value for each bin.
7. The method of claim 1 where the variance, σ 2 , is determined by:
σ
2
=
∑
i
=
1
N
f
norm
(
i
)
·
(
x
(
i
)
-
μ
)
2
where:
N is a number of bins,
f norm (i) is a normalized density function for each bin,
x(i) is a center intensity value for each bin.
8. A method for detecting an object on a road in an image taken in a vehicle, the method comprising:
acquiring the image;
determining a gradient magnitude and angle for each pixel of the image;
applying a double threshold to the image with an upper and a lower threshold for determining edges of the object, where the upper and the lower threshold are determined by a method comprising:
determining a histogram of the image;
dividing the histogram into a predetermined number of bins;
determining a center value for each bin;
determining a normalized density function for, each bin;
determining a normalized probability density function from the normalized density function for each bin;
determining a mean value and a variance of the normalized probability density function; and
determining the upper and the lower threshold on the basis of the mean value and the variance by using the following equations:
T 1 =max(0,(μ−σ)/ a )
T 2 =min( A, (μ+σ)/ b ) where:
μ is the mean value of the normalized probability density function,
σ is the variance of the probability density function,
a is a parameter between 5 and 7, preferably 6,
b is a parameter between 2 and 4, preferably 3, and
A is the maximum intensity value.
9. The method of claim 1 further comprising the step of determining a center intensity value for each bin.
10. The method of claim 1 further comprising the step of determining the number of pixels contained in each bin, where the normalized density function is determined on the basis of the number of pixels in each bin and on the basis of the number of pixels in the image.
11. The method of claim 5 where the mean value is determined on the basis of the normalized density function of each bin and on the basis of the center intensity value for each bin.
12. The method of claim 1 where;
the edges are detected in a series of images, and
the upper and the lower threshold are calculated automatically for, each image based on the distribution of intensity values of the corresponding image.