Research And Implementation Of Infrared Object Detection Algorithm Based On Online Learning

The object detection based on computer vision has become a popular research topic in the world.However,there are still some challenges in the research of object detection.In the last few years,with the development of intelligent video monitor,smart city and intelligent vehicle driver system,the demand of object detection technology is higher.In addition,the mature infrared imaging technology has also promoted the technology of infrared object detection.This thesis focuses on the discussion of infrared object detection in feature extraction and online learning technology,introduces several characteristics,and puts forward a kind of improved feature descriptor through the comparison and analysis of pedestrian detection experiment.In the process of infrared pedestrian detection,the thesis uses this improved descriptor for feature extraction,and updates classifiers by continuously learning the the characteristics new samples online,in order to improve the capability of object detection system.There are two main content parts,namely,the object feature extraction and the online learning.First of all,based on the traditional LBP feature,with expansion of filter banks,the thesis puts forward an improved local two value model descriptors,FBLBP descriptor.Using FBLBP descriptor describes the infrared object in the detection system,and extracts FBLBP feature for for online learning training of classifiers.In the third chapter,the FBLBP descriptor is proved validity and feasibility through lots of experiments.Then,the thesis uses a Triple Online Boosting Training online learning algorithm for online learning object features,through continuously learning the sample information in different scenarios,improves the detection ability and classification ability of the classifier.At the same time,in the process of online learning,the collaborative training algorithm based on Tri-training is used to automatically label the results of the test.Finally,thesis concludes and analyses the work,and discusses the deficiencies,and also makes a further research plan.