Method for automatically obtaining outdoor scene text in video based on feature summary diagram

Abstract

The invention discloses a method for automatically obtaining an outdoor scene text in a video based on a feature summary diagram. The method comprises the following steps that: firstly, obtaining thevideo frame image of a scene text, and generating a video frame feature summary diagram on the basis of the RGB (Red, Green and Blue) color space of a video frame image: firstly, independently extracting four convolution diagrams in four directions including a horizontal direction, a vertical direction, a 45-degree direction and a 135-degree direction, obtaining four-direction feature vectors which represent the color space so as to obtain ten saliency maps which represent the video frames in different directions to obtain the video frame feature summary diagram; and then, on the basis of thevideo frame feature summary diagram and the RGB color space, carrying out K mean value color clustering calculation, independently analyzing the connected domains of four categories of results after four categories of results which shows background and foreground word characters, character outlines and noise, deleting two areas of background and noise, and automatically obtaining a final outdoor scene text. The method has the advantages of simple step, easiness in calculation and good popularization and application prospect, and the outdoor scene text can be obtained and identified in real time.

Description

technical field [0001] The invention relates to a digital image processing method, in particular to a method for automatically acquiring outdoor scene text in a video based on a feature summary graph, and belongs to the technical field of computer vision processing. Background technique [0002] In the past few years, with the popularization of digital image acquisition devices, smart phones and practical vision systems and their equipment, content-based image understanding technology has received more and more attention. Because the scene text in the image / video has relatively rich and direct semantic information clues, the scene text is considered as an important object that must be detected and recognized. Among them, text detection, location, extraction and recognition are the main steps to obtain text information. Usually, the operations of text detection, location and extraction are collectively referred to as text acquisition. For text recognition, text acquisition ...

AI Technical Summary

Problems solved by technology

The disadvantage is: because it belongs to the global calculation method, it is sensitive to non-uniform illumination, and the calculation cost and the selection of parameter k when analyzing multiple candidate images are very complicated

However, the algorithm is sensitive to the size of extracted characters and is not suitable for extracting scene text in videos

[0013] In short, there are many unsatisfactory aspects in the detection and location of the scene text and the extraction technology of the scene text in the above-mentioned various existing technologies. Therefore, how to develop a scene text in a video with relatively good performance or complete features The acquisition method has become a new topic of great concern to the technical personnel in the industry

Images