Vectorization and Optimization of Fog Removal Algorithm


A portion of the picture handling calculations are verycostly regarding activities and time. To utilize these algorithmsin constant condition, streamlining and vectorization arenecessary. In this paper, approaches are proposed to enhance, vectorize and how to fit the calculation in low memory space. Here, advanced anisotropic dissemination based haze expulsion calculation isproposed. Mist expulsion calculation expels the haze from imageand produces a picture having better perceivability.

This algorithmhas numerous stages like anisotropic dissemination, histogram stretchingand smoothing. Anisotropic dissemination is an iterative procedure thattakes about 70% of time unpredictability of the entire calculation. Here, improvement and vectorization of the anisotropic dispersion is proposed for better execution. Nonetheless, optimizationtechniques cost some exactness however that can be disregarded forsignificant change in execution.

For memory constraintenvironment, a technique is proposed to process the whole blockof picture and keeps up the uprightness of tasks. Resultsconfirm that with our enhancement and vectorization approaches, execution is expanded up to 90 fps (roughly) for VGAimage on one of the picture preparing DSP test system. Regardless of whether, framework doesn’t have vector tasks, the proposed optimizationtechniques can be utilized to accomplish better execution (2× quicker).

BASE PAPER: Vectorization and optimization

Leave a Reply

Your email address will not be published. Required fields are marked *