The digital video stabilization is oriented on the removal of unintentional motions from video sequences caused by camera vibrations under external conditions, motion of robots stabilized platforms in a rugged landscape, a sea, oceans, or jitters during a non-professional hand-held shooting. The approaches for digital video stabilization in static and dynamic scenes are similar. However, objectively the analysis of dynamic scenes is needed in advanced intelligent methods. Several sequential stages include the choice of the key frames, the local and global motion estimations, the jitters compensation algorithm, the inpainting of frames boundaries, and the blurred frames restoration, for which the novel methods and algorithms were developed. The proposed application of fuzzy logic operators improves the separation results between the unwanted motion and the real motion of rigid objects. The corrective algorithm compensates the unwanted motion in frames; thereby the scene is aligned. The quality of stabilization in test video sequences was estimated by Peak Signal to Noise Ratio (PSNR) and Interframe Transformation Fidelity (ITF) metrics. During experiments, the PSNR and ITF estimations were received for six video sequences received from the static camera and eight video sequences received from the moving camera. The ITF estimations increase up on 3–4 dB or 15–20% relative to the original video sequences.
|Title of host publication||Computer Vision in Control Systems-1|
|Subtitle of host publication||Mathematical Theory|
|Editors||Margarita N. Favorskaya, Lakhmi C. Jain|
|Place of Publication||New York|
|Number of pages||49|
|Publication status||Published - 2015|
|Name||Intelligent Systems Reference Library|