Multi Traffic Scene Detection based on Machine Learning Algorithm

Abstract

The problem of removing one type of natural noise such as rain from the given rain image is been discussed here. The performance of computer vision and Image analysis process can be gravely affected by this rain streaks problem which damaged the visibility of an image and also it develops interference in it. So, in order to get a streak free background vision, we are in need to develop algorithms which helps us to remove streaks from an image called Rain Streak Removal Algorithms. This problem is formulated as a layer decomposition which enables to view the true background scene content implosively without any streaks. The appearance of rain streak structure is infelicity of low rank in our existing system and also it is derived using a Sparse dictionary learning method. Even though it provides a quite clear vision to see, their practical performance is unsatisfied. For that, we are proposing a another method to retrieve a image content without any stripes. The prior method is learned on small patches based on the model of Gaussian Mixture. In addition to this, we create a step to separate the residues of background image to recover the structure for it and also it improves the performance quality of decomposition. Then the evaluation process takes place to categorize the implosive performance of quantitative structure than the existing systems in terms of large margin.