AUTOMATIC LOW CONTRAST DETECTION

Number: 20220156978
Published: 2022-05-19
Filed: 2020-11-18
Assignee: Disney Enterprises, Inc.
Inventors: FARRÉ GUIU; Miquel Angel, JUNYENT MARTIN; Marc, PERNIAS; Pablo
CPC: G06V10/751; G06V10/40; G06V10/56; G06T7/90; G06V10/44
Verdict: Set aside generic image QC (contrast detection)
Source: Google Patents · FreePatentsOnline

Abstract

A method includes generating a delicate area map by performing a morphological function on a portion of a received first image and identifying a plurality of edges in the first image, the plurality of edges comprising a plurality of pixels. The method also includes verifying a first contrast metric for a first subset of pixels that are in the plurality of pixels but not in the delicate area map, verifying a second contrast metric for a second subset of pixels that are in the plurality of pixels and in the delicate area map, and generating a validation result based on the verifying of the first contrast metric and the verifying of the second contrast metric.

Background

BACKGROUND

Images contain pixels that may have high or low contrast. Low contrast pixels have a color that is similar to the colors of its surrounding pixels, and high contrast pixels have a color that is different from the colors of its surrounding pixels. It may be difficult for the human eye to differentiate a low contrast pixel from its surrounding pixels. On the other hand, it is relatively easy for the human eye to differentiate a high contrast pixel from its surrounding pixels. Therefore, when the creator of an image wants an object in the image to stand out and be easily detectable by the human eye, the creator will want that object to have a high contrast with its surroundings.

Conventional methods of detecting contrast are subjective and rely on human judgment or expertise. As a result, these conventional methods may produce inconsistent results. For example, one reviewer may determine that a region of an image has low contrast while another reviewer may determine that that same region has high contrast. Additionally, these conventional methods are subject to human error and their results may vary depending on the media on which the image is displayed (e.g., mobile phone display vs office projector). For example, a reviewer who is identifying regions of low contrast may miss or not see a low contrast region due to its small size.

Claims

1. A method comprising: generating a delicate area map by performing a morphological function on a portion of a received first image; identifying a plurality of edges in the first image, the plurality of edges comprising a plurality of pixels; verifying a first contrast metric for a first subset of pixels that are in the plurality of pixels but not in the delicate area map; verifying a second contrast metric for a second subset of pixels that are in the plurality of pixels and in the delicate area map; and generating a validation result based on the verifying of the first contrast metric and the verifying of the second contrast metric. || 14. A system comprising: a memory; and a hardware processor communicatively coupled to the memory, the hardware processor configured to: generate a delicate area map by performing a morphological function on a portion of a received first image; identify a plurality of edges in the first image, the plurality of edges comprising a plurality of pixels; verify a first contrast metric for a first subset of pixels that are in the plurality of pixels but not in the delicate area map; verify a second contrast metric for a second subset of pixels that are in the plurality of pixels and in the delicate area map; and generate a validation result based on the verifying of the first contrast metric and the verifying of the second contrast metric.