is and for two frames each. Then we can
haveTeX Source
So we can get the difference between two frames:
TeX Source
Use Canny edge detection with frames get the result:
, We can get edge image , Then we can
TeX Source
are edge images caused by
Define signal noise ratio is
and each.
TeX Source
While is the number of edges caused by moving objects, and is the
number of edges caused by noise.
Then we know the SNR of the moving edge method is
TeX Source
2. In our method, we first get edge images by Canny edge
detector:
Then by difference we get
TeX Source
TeX Source
Since in the practical system, the difference between two edge images is absolute value of the difference value and the edges of two images are not the same when the objects are moving.
TeX Source
So actually in the edge difference image we can have the sum of the edges of two frames.
TeX Source
Because the noise is independent and two frames are dependent with each other, we can have
TeX Source
The SNR in our algorithm is
TeX Source
It shows that the SNR in our algorithm is less than the moving edge method. Our method will work more efficiently.
3.3 Object locationing
After separating moving objects and background, we need to locate the object so as to get the exact position of moving objects. The common approach is to calculate connected components in binary images, delete those connected components whose area are so small, and get circum-rectangle of the object.
One of the most familiar connected components algorithms is a pixel labeled algorithm based on eight conjoint areas. According to this approach, zero is assigned to each pixel of background and an only number we called tab is assigned to every connected component while the biggest tab represents the number of connected components in the image. Basic steps of this algorithm are as follows:
Travel the entire image from top to bottom and from left to right while a non-zero value
, The value is selected according to the tab of
is assigned to each non-zero pixel conjoint areas. ?
If all the conjoint areas are part of the background which means the value of the pixels of all these areas are zero, a new tab which is never used will be assigned to
.
? If there is only one conjoint area has a nonzero tab, this tab will be assigned to .
? If there are more than one conjoint areas whose tab are not zero and some of these tabs are different, the smallest one among these tabs will be assigned to
while all
the other pixels whose tab is same as one of the rest of these tabs will be changed to the smallest one as well.
As we can see from the steps mentioned above, the time complexity of connected components labeling is
, while
and
represent the height and width of the image. Our
approach divides the entire image into separate blocks during object locating before connected components labeling so as to diminish the amount of operation evidently. If the size of original image is
and we divide the image into
small blocks to
be operated, the time of operation can be diminished to 1/1200 of the original one. It improves the speed of detection greatly, and the result of the latter experiment validates this conclusion as well.
EXPERIMENTAL RESULTS AND ANALYSIS
In this paper, an improved moving object detection algorithm based on frame difference and edge detection is brought forward. It was simulated on the PC with Pentium 4 CPU and 2.50GHz 256 EMS memory. The operating environment is Windows XP. Programming environment is Matlab 7.0. Size of the sequence image is results are as follows.
. Partial stimulation
Figure 4 Three continuous frames Previous | View All | Next
Figure 5 Detection results by using the improved algorithm Previous | View All
This paper also achieves several classical algorithms (frame difference method [5], moving edge method [6], background subtraction method [7]), and compared their performance with our improved algorithm by detecting moving object from the video sequences captured in the laboratory which has 250 frames and 625 moving objects totally. The analysis results are shown in the following table.