Method and device for text detection

Abstract

The invention provides a method and device for text detection which are used for positioning a text region in a natural scene image. The disclosure relates to a text detection method and device and a text information extraction method and system. The method used for detecting a text region in an image comprises the steps of generating assemblies according to an input image; based on the similarity between feature values of any two assemblies, generating an assembly connector used for connecting the two assemblies; determining a main direction in allusion to each assembly connected by the assembly connectors to act as a main direction of each assembly according to direction distribution of the assembly connectors connected to each assembly; dividing the assemblies connected by the assembly connectors into at least one assembly group based on the difference between the main direction of the assemblies and a direction of the assembly connector used for connecting the assemblies; and generating a text region based on the at least one assembly group. Compared with the prior art, the method and device provided by the invention can position a text region in an image accurately, and the detection result can be further used for text recognition.

Description

technical field [0001] The invention relates to computer vision, image processing and understanding, in particular to a multi-directional text detection method and device. Background technique [0002] Text in natural scene images or video frames carries important information for visual content understanding and retrieval. Extracting textual information or detecting text in images, especially in natural images or video frames, is critical for many computer vision applications (e.g., computerized assistance for visually impaired persons or foreigners, automatic retrieval of images or videos, and urban environments). It is crucial for robot navigation in , etc.). However, text retrieval in natural scenes is a difficult problem, and the main challenge lies in diverse texts with different fonts, sizes, skew angles or distortions, etc. Environmental factors like uneven illumination and reflections, poorly lit conditions, and complex backgrounds add further complexity. [0003]...

AI Technical Summary

Problems solved by technology

[0007] However, in the case where the layout of the text is complex, or in the case where multiple text lines with different directions are located in a local area, the methods disclosed above can only detect text in a specific direction, making the text in different text lines Components are incorrectly grouped into the same group and lead to false detection results

Images