Detecting Motion in Single Images Using ALISA

Motionanalysis often relies on differencing operations that inherently amplify noise and are hindered by the spatial correspondenceproblem.Analternative approach is proposedusing AIJSA(Adaptive learning Imageand SignalAnalysis)to detect differences in types of motion by classifying the imagingeffects of the motion in single frames.Withan appropriatefeature set, the ALISA engine accumulatesa multi-dimensionalhistogram that estimates the probability density function of a feature spaceandusesthe result as a basis for classification. Asa functi(m of imagesamplingrate and the scale of image structures, the AUSAengine wasable to discriminate betweena slow movingand fast movingobject with a confidencegreater than 99%.

• Type: Conference
• Year: 1998
• Where: FLAIRS
• Authors: Taras P. Riopka, Peter Bock
• Keywords: Artificial Intelligence, FLAIRS 1998, Probability Density Function, SignalAnalysis)to Detect Differences, Spatial Correspondenceproblem.Analternative Approach