Digital Signal Processing Reference
In-Depth Information
Fig. 4.
Flow chart for contour extraction algorithm
(a) (b) (c) (g) (h) (i)
(d) (e) (f) (j) (k)
Fig. 5.
Shows the (a)visible and (d)infrared images, and the (g)fused image, the second column
shows the results by frame differential and the third column shows the results by GMM,
(j)shows the temp result before hole filling and (k) shows the final result.
image only, as (e, f) shows, the halo effect and thermal noise are the stumbling blocks
to achieve a complete and clear detection. The fusion-technology helps integrating the
characteristics of both visual and infrared image, is favorable to moving target extrac-
tion. Image noise is significantly reduced in the fused image (shown in h, i), some
incomplete parts will be improved by contour detection as a subsequent patches. More
