ABSTRACT
In the machine learning world, moving object detection is a very interesting problem for researchers and many researchers are working on it for getting optimal results. It is very difficult to get the accurate results in this problem area due to very large diversity in the set of color and shapes of the object. Several techniques are already being used by researchers. In this chapter, machine learning algorithms are used for motion detection and consecutive images. Movement: the pixel is a useful characteristic of image for the segmentation process as it distinguishes a moving pixel from the motionless pixels. Modern machine learning techniques give better results in classification type of problems. Machine learning can be used to identify moving pixels using the successful trained model. Random Forest Classifier is used to classify the pixel in two classes of moving and stable pixels. Results obtained using this approach show that machine learning methods are very promising techniques for this type of problem area. Random Forest Classifier works better with the numerical features set. It enhances the result up to 6 percent over SVM classifier for all sequences with 5 percent improvement over LS-SVM for the sequences having the stable background. It can handle large size input as well.
