A Novel Adaboost Regression Classifier for Video Retrieval in Video Sequence
This paper presents a new method for video retrieval, based on machine learning with regression. The proposed classification technique integrates Adaboost and regression classifier for significant retrieval of video frame. The proposed method consists of three stages such as key frames segmentation and gradient of pixels. In this technique, Adaboost classifier is involved in removal of noisy or blurred pixel of the segmented frame. Regression technique converts the video frame pixel either 0’s or 1’s which eliminates the noises in the frame. For the query video, the adopted classifier evaluates the machine learning system for retrieval of similar frames in the databases using proposed Adaboost Regression (ABR) classifier. Experimental analysis is conducted for video datasets to evaluate the proposed ABR classifier performance evaluation. Results stated that through proposed ABR approach incorporated in machine learning system effectively retrieve video frame for query frame. The proposed ABR classifier technique significantly improves the retrieval rate in terms of accuracy, precision.