Case study on wavelet choice based on statistical image quality measures

摘要

Surveillance applications are under many negative influences, which should be suppressed, because these negative influences result in incorrectness of the motion mask. Suppression of several conflicting requirements can be optimized by a multiobjective approach. This paper proposes a multiobjective approach to selection of wavelets based on two main objectives: statistical image quality measures and execution time. Execution time is a measure of wavelets complexity. Segmentation is the final goal in order to insure precise operation of any surveillance algorithm. This paper presents a case study considering one, two, three, and four goals for wavelet selection comparison. Different wavelets are found to be an optimal choice for different weights of the objectives.