A flame detection algorithm based on Bag-of-Features in the YUV color space

摘要: Computer vision-based fire detection involves flame and smoke detection. This paper proposes a new algorithm, which is based on Bag-of-Features technique in the YUV color space. Inspired by that of image video will fall certain regions space, models pixels non-flame are established code book training phase our proposal. In testing phase, input split into some N×N blocks each block classified respectively. block, values space extracted as features, just phase. According to experimental results, proposed method can reduce number false alarms greatly compared with an alternative while it also ensures accurate classification positive samples. The performance better than algorithms.