New Saliency Point Detection and Evaluation Methods for Finding Structural Differences in Remote Sensing Images of Long Time-Span Samples - Advanced Concepts for Intelligent Vision Systems - page 273

contours; A graph based shape descriptor will be shown based on the saliency
points of the difference and in-layer features; finally, we prove the methods ca-
pabilities for finding structural changes on remote sensing images.

2 Change Detection with Harris Keypoints

2.1 Harris Corner Detector

The detector was introduced by Chris Harris and Mike Stephens in 1988 [15]. The
algorithm based on the principle that at corner points intensity values change
largely in multiple directions. By considering a local window in the image and
determining the average changes of image intensity result from shifting the win-
dow by a small amount in various directions, all the shifts will result in large
change in case of a corner point. Thus corner can be detected by finding when
the minimum change produced by any of shifts is large.
The method first computes the Harris matrix ( M ) for each pixel in the image.
Then, instead of computing the eigenvalues of M ,an R corner response is defined:

Tr 2 ( M ) R = Det ( M ) − k ∗ (1)

This R charasteristic function is used to detect corners. R is large and positive
in corner regions, and negative in edge regions. By searching for local maximas
of a normalized R , the Harris keypoints can be found. Normalizing makes R
smoother and only major corner points are detected. R could also be used for
edge detection:

function is large and positive in corner and also positive but
smaller in edge regions, and nearly zero in flat regions. We used this function
in our later work. Figure 1 shows the result of Harris keypoint detection. On
Figure 1(b) light regions shows the larger R values, so keypoints will be detected
in these areas (Figure 1(c)).

| R |

2.2 Change Detection

The advantage of Harris detector is its strong invariance to rotation and the
R characteristic function's invariance to illumination variation and image noise.
Therefore it could be used eciently for change detection in airborne images. In

(a) Original (b) R function (c) Keypoints

Fig. 1. Operation of Harris detector: Corner points are chosen as the local maximas of
the R characteristic function