Image processing methods and apparatuses to obtain a narrow depth-of-field image

摘要

Disclosed are example methods and apparatuses for image processing for improving an out-of-focus effect. A disclosed example methods includes capturing a first image, a second image, and a third image, which are captured with a different focus for a same subject and background; setting a subject portion and a background portion using the first and second images; combining the subject portion of first image and the background portion of the third image with each other to obtain an narrow depth-of-field (DOF) image; and performing image processing on the obtained narrow DOF image.

主权项

1. An image processing method, the method comprising:
capturing a first image, a second image, and a third image, which are captured with a different focus for a same subject and background; setting a subject portion and a background portion using the first and second images: combining the subject portion of the first image and the background portion of the third image with each other to obtain an narrow depth-of-field (DOF) image; performing image processing on the obtained narrow DOF image; wherein the setting of the subject portion and the background portion comprises: generating a first difference of Gaussian (DoG) image and a second DoG image by applying a DoG filter to each of the first and second images; comparing radio-frequency components in mask regions of the first and second DoG images with each other, wherein the mask regions respectively include reference pixels that are located at corresponding positions of the first and second DoG images; if, as a result of the comparing, it is determined that radio-frequency components of the first DoG image are greater than those of the second DoG image, the reference pixels are represented as a first brightness value, and if it is determined that radio-components of the second DoG image are greater than those of the first DoG image, the reference pixels are represented as a second brightness value, to derive a single binary-coded image; and correcting error pixels included in the derived binary-coded image.