Video stabilization: digital and mechanical approaches

Download
2008
Bayrak, Serhat
General video stabilization techniques which are digital, mechanical and optical are discussed. Under the concept of video stabilization, various digital motion estimation and motion correction algorithms are implemented. For motion estimation, in addition to digital approach, a mechanical approach is implemented also. Then all implemented motion estimation and motion correction algorithms are compared with respect to their computational times and accuracies over various videos. For small amount of jitter, digital motion estimation performs well in real time. But for big amount of motion, digital motion estimation takes very long time so for these cases mechanical motion estimation is preferred due to its speed in estimation although digital motion estimation performs better. Thus, when mechanical motion estimation is used first and then this estimate is used as the initial estimate for digital motion estimation, the same accuracy as digital estimation is obtained in approximately the same time as mechanical estimation. For motion correction Kalman and Fuzzy filtering perform better than lowpass and moving average filtering.

Suggestions

3d geometric hashing using transform invariant features
Eskizara, Ömer; Ulusoy, İlkay; Department of Electrical and Electronics Engineering (2009)
3D object recognition is performed by using geometric hashing where transformation and scale invariant 3D surface features are utilized. 3D features are extracted from object surfaces after a scale space search where size of each feature is also estimated. Scale space is constructed based on orientation invariant surface curvature values which classify each surface point's shape. Extracted features are grouped into triplets and orientation invariant descriptors are defined for each triplet. Each pose of eac...
Single and dual band quantum well infrared photodetector focal plane arrays on inp substrates
Eker, Süleyman Umut; Beşikci, Cengiz; Department of Electrical and Electronics Engineering (2010)
Excellent uniformity and mature material properties of Quantum Well Infrared Photodetectors (QWIPs) have allowed the realization of large format, low cost staring focal plane arrays (FPAs) in various thermal imaging bands. AlGaAs/InGaAs and AlGaAs/GaAs materials systems have been the standard systems for the construction of mid-wavelength infrared (MWIR) and long-wavelength (LWIR) QWIPs. However AlGaAs/GaAs QWIP FPAs suffer from low quantum and conversion efficiencies under high frame rate (low integration ...
FPGA implementation of real time digital video stabilization
Özsaraç, İsmail; Ulusoy, İsmail; Department of Electrical and Electronics Engineering (2011)
Video stabilization methods are classified as mechanical and digital. Mechanical methods are based on motion sensors. Digital methods are computer programs and classified into two as time domain and frequency domain based on the signal processing methods used for the motion analysis. Although, mechanical methods have good real time stabilization performance, they are not suitable for small platforms such as mobile robots. On the other hand, digital video stabilization methods are easy to implement on variou...
Capacitive cmos readout circuits for high performance mems accelerometers
Kepenek, Reha; Külah, Haluk; Department of Electrical and Electronics Engineering (2008)
This thesis presents the development of high resolution, wide dynamic range sigma-delta type readout circuits for capacitive MEMS accelerometers. Designed readout circuit employs fully differential closed loop structure with digital output, achieving high oversampling ratio and high resolution. The simulations of the readout circuit together with the accelerometer sensor are performed using the models constructed in Cadence and Matlab Simulink environments. The simulations verified the stability and proper ...
Traffic sign recognition for unmanned vehicle control
Havur, Mehmet Bülent; Bulut, Mehmet Mete; Department of Electrical and Electronics Engineering (2006)
In this thesis, video frames acquired by a camera in a moving car are processed for detection of candidates of triangular, rectangular and circular traffic/road signs based on mainly shape information by performing contour analysis. Color information is utilized as an auxiliary method to improve detection. Then recognition based on template matching is realized on detected traffic/road sign candidates. Detection and recognition results of traffic/road signs in video frames taken in different time intervals ...
Citation Formats
S. Bayrak, “Video stabilization: digital and mechanical approaches,” M.S. - Master of Science, Middle East Technical University, 2008.