Detection Arithmetic Of The Title In TV Image

Nowadays, the rapid growth of television programs has brought heavy workload to the TV station technicians, for bringing stress and energy-consuming tasks forward. Some are required to the 24x7 continuous watches on the TV screen, others needs to review material of video image without stopping. From the standpoint of the TV station, it means a great expenditure on manpower and other resources. In views of the above, it will be a valuable study on developing effective techniques to manage video materials and improving the efficiency of video surveillances. In addition, the related techniques can also be applied to other industries. For examples, car plate recognitions, multi-media resource management, data recording for live matches and etc, which share the same need for text detection and recognition of multi-media images.What's happening in the TV station is that a lot tasks require massive manual work on subtitle detection and processing. Moreover, most of them are relatively inefficient which could cause a waste of resources. This article is based on the actual needs, aiming to acquire a highly efficient subtitle detections supported by the algorithm introduced. Firstly I'll brief the status quo of the researches in this field and the technologies used at home and abroad. Secondly, give analysis and summarize the specialties of the TV subtitles in China. According to these characteristics, and put forward the algorithm based on edge detection algorithm firstly extracted features, and then based on text captions and distribution characteristics of judging marquee area, Afterwards, separate subtitle from the background image through subtitle gray value and color information. At last, we archive the features expected and an encouraging result.