Character area extracting device, image picking-up device provided with character area extracting function and character area extracting program

Abstract

In order to accurately specify a character region from image data including portions with different gradation contrasts, the character region extracting device receives image data obtained by photographing a subject, binarizes the entire image data using a first threshold, and Separation into reflective areas where the number of pixels is saturated and non-reflective areas where the number of pixels is not saturated. Next, the reflection area is binarized and separated into a character area and a background area by using a second threshold. Similarly, the non-reflection area is binarized and separated into a character area and a background area by using the third threshold. The respective character areas of the reflective area and the non-reflective area are combined, and position information of the character areas in the image data is determined.

Description

technical field [0001] The present invention relates to the process of extracting character regions from image data. Specifically, it relates to obtaining binary image data to be subjected to character recognition processing from image data obtained by photographing a subject, and to extract with high precision even from image data that is reflected due to ambient light mapping. Handling of character regions. [0002] The character area is a pixel area classified into character pixels. One or more character regions arranged on the same baseline are called "rows". Background technique [0003] Today, as compact digital cameras and mobile phones equipped with camera functions have spread, high-precision character recognition functions are being pursued for camera photography functions. [0004] The target image data of the conventional character recognition processing is data obtained in a non-interfering shooting environment because a subject is captured by a fixed device ...

AI Technical Summary

Problems solved by technology

[0012] However, in images with reflective areas, in addition to the borders of characters and backgrounds, edges are also produced at the boundaries of reflective and non-reflective areas

Therefore, in Patent Document 1, it is considered that the edges of the reflective area and the non-reflective area generated on the background area are sometimes erroneously detected as part of the character area, resulting in a decrease in the accuracy of character area extraction, which in turn leads to Recognition Accuracy Drops in Character Recognition Processing

[0013] In addition, in Patent Document 2, it is considered that a reasonable threshold value for separation processing cannot be detected on the histogram when the shading values ​​of pixels belonging to the character area and the background area are not uniform in each area.

In addition, since the evaluation of the threshold value and the correction based on the evaluation are not performed, the separation accuracy of the character area and the background area decreases, resulting in a decrease in the extraction accuracy of the character area, which in turn leads to a decrease in the recognition accuracy in the character recognition process.

Images