Text image segmentation method and device

Abstract

The invention discloses a text image segmentation method and device. The method includes the following steps that: a text image is acquired, and the text image is initially divided into a plurality ofimage blocks according to priori rules; internal points in each image block are determined, and blank pixels inside characters in each image block are eliminated through filling the internal points;the average connected domain value of each image block is calculated through a projection mode; connected domain expansion is performed on each image block according to the average connected domain value; and the edges of character areas are extracted from each image block, and each image block is divided according to the edges. With the text image segmentation method and device adopted, differenttext images or different types of text images can be segmented. The accuracy and universality of the method and device can achieve satisfactory levels.

Description

technical field [0001] The present invention relates to the field of computer security, and more specifically, to a text image segmentation method and device. Background technique [0002] With the popularity of machine learning, artificial intelligence and other fields, various fields related to it will inevitably be driven up. As we all know, artificial intelligence is based on machine learning, and for machine learning, it is first necessary to store historical data in the computer. Then, these data are processed through machine learning algorithms. This process is called "training" in machine learning, and the processed results can be used to predict new data. This result is generally called a "model". The process of making predictions on new data is called "prediction" in machine learning. "Training" and "prediction" are two processes of machine learning, "model" is the intermediate output of the process, "training" produces "model", and "model" guides "prediction". ...

AI Technical Summary

Problems solved by technology

The blank background method separates each module of the foreground area in the picture. This method is simple and suitable for layout types with large blank areas, but the disadvantage is that the blank areas of the layout are required to be connected, and the adaptability is poor. For some complex layouts, Poor processing effect

The connected domain method starts from the image as a whole, and obtains some basic parameters, such as word spacing, connected threshold, etc., through the projection operation on the image pixels, and connects the image through the connected threshold to obtain the entire text block for segmentation. The advantage of the method is that it has strong adaptability to various layouts, and the time complexity of the algorithm is also relatively good. The disadvantage is that the segmentation effect is poor for some layouts where the connectivity threshold feature is not obvious.

For the unsatisfactory problems in the traditional image and text recognition methods in the prior art, there is no effective solution yet

Images