Abstract- Video object co-segmentation is a method of segmenting and discovering primary object from multiple videos without human annotation. In this paper, we are presenting a firefly based video object co-segmentation method. Here we are using SIFT flow method for object segmentation and firefly algorithm for object optimization. SIFT flow descriptor discovers the corresponding object using inter-frame motion flow from optical flow. But it will not capture the optimal motion of inter-frame. So we are using Firefly approach in addition to SIFT flow. Firefly approach captures optimal inter-frame motion based on the velocity updation and position of the particle for better results. SIFT flow produces segmented primary objects over the entire video data set. Finally we combine this segmented object to get high quality original videos. The results that we obtained from conducting the experiment show that the Firefly based co-segmentation method achieves high accuracy compared with the existing co-segmentation method.
Keywords— Video object co-segmentation, energy optimization, object refinement, spatio-temporal scale-invariant feature transform (SIFT) flow.
AD Publications is a rapidly growing academic publisher in the fields of Engineering, Medical-Health, Environmental Science and Agriculture Research. AD Publications is a registered organization broad-based open access and publishes most exciting researches with respect to the subjects of our journals. The Journals is being indexed and abstracted by all major global current awareness and alerting services.
The organization aims at undertaking, co- coordinating and promoting research and development. It provides professional and academic guidance in the field of basic education, Higher Education as well in the Technical Education. Our Aims is to Promote and support, High Quality basic, Scientific Research and development in fields of Engineering, Medical-Health, Environmental Science and Agriculture Research and to Generate Public awareness, provide advice to scholar’s researchers and communicate research outcomes.