SEARCH
0-9 A B C D E F G H I J K L M N O P Q R S T U V W X Y Z
Prev | Current Page 256 | Next

Yingshu Li, My T. Thai, and Weili Wu

"Wireless Sensor Networks and Applications"

However, with a large
threshold, the missed detection rate will also increase. Similarly, having a large
probing area can also improve the performance since the statistical analysis
can be more accurate if more values of event predicates can be collected.
But the communication overhead for a large probing area will also increase
significantly
Thus, an interesting topic here would be the selection of the threshold
value t0 and the size of probing radius R. To reach a satisfactory operation,
one can pre-calculate the performance curves and store the selected threshold
and probing radius into sensors such that the statistical approach can satisfy
the preferred performance criteria.
152
Chapter 6 Boundary Detection for Sensor Networks
2.2 Image processing based approach
A number of edge detection schemes have been designed and analyzed in the
image processing literature. A typical method is to make use of a high pass
filter (e.g., Prewitt, Sobel filters) to detect edges in an image. By convoluting
an image of interest with a high pass filter, regions of high spatial gradient
that correspond to edges will be emphasized.
Usually, a high pass filter G(x, y) consists of at least a pair of 3x3 convolution
masks, one mask for each of the two orthogonal axes. For example,
the Prewitt edge detector defines mask Gx to calculate the gradient along the
horizontal orientation, and Gy is used to estimate the gradient of the vertical
orientation.


Pages:
244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268