Color-based writing trace extraction method, system and device under complex background

Abstract

The invention belongs to the technical field of target extraction, particularly relates to a color-based writing trace extraction method, system and device under a complex background, and aims to solve the problems that an existing extraction method is incomplete in writing trace extraction and high in image quality requirement, and writing traces with changed colors cannot be extracted under a complex environment. The method comprises the steps of converting a to-be-detected image containing a to-be-detected writing trace into an HSV color space, selecting seed points based on the HSV color space, and dynamically obtaining pixel points on an edge image from the seed points to serve as newly-added to-be-traversed points; and based on the seed points and newly added points to be traversed,obtaining a pixel set of writing traces to be detected through image refinement. According to the method, the connectivity of the correction trace can be enhanced, and the correction trace can be effectively extracted in a complex scene; meanwhile, the conditions of newly-added points to be traversed are strictly set, the outline of the writing trace can be effectively controlled, excessive noiseis prevented from being introduced, and subsequent processing is facilitated.

Description

technical field [0001] The invention belongs to the technical field of target extraction, and in particular relates to a color-based writing trace extraction method, system and device under complex backgrounds. Background technique [0002] To achieve the target extraction task through computer image processing, the current very popular method is to use the deep learning method. With the support of excellent data sets, the trained deep model can well complete common instance segmentation tasks. However, writing traces are different from large targets and large objects. The lines in writing traces are slender, changeable, and do not have texture information, and the edge information is easily disturbed by the surroundings. Therefore, the commonly used convolution operation is used in the detection and feature extraction of writing traces. Performance is not good. [0003] In the prior art, target extraction based on the HSV color space can realize the extraction of large tar...

AI Technical Summary

Problems solved by technology

[0006] In order to solve the above problems in the prior art, that is, in order to solve the problem that the existing target extraction method is incomplete in the extraction of writing traces, has high requirements on image quality, and cannot extract writing traces with changed colors in complex environments, the present invention proposes A method for extracting writing traces based on color under complex backgrounds is proposed, which includes:

Images