Detecting glare in a frame of image data

Patent metadata

Field	Value
Publication number	US-2017039684-A1
Application number	US-201615228458-A
Country	US
Kind code	A1
Filing date	Aug 4, 2016
Priority date	Dec 9, 2008
Publication date	Feb 9, 2017
Grant date	—

Abstract

Disclosed are systems, computer-readable mediums, and methods for detecting glare in a frame of image data. A frame of image data is preprocessed. A set of connected components in the preprocessed frame is determined. A set of statistics is calculated for one or more connected components in the set of connected components. A decision for the one or more connected components is made, using the calculated set of statistics, if the connected component is a light spot over text. Whether glare is present in the frame is determined.

First claim

What is claimed is:

1 . A method for detecting glare in a frame of image data, the method comprising:
preprocessing the frame of the image data; determining a set of connected components in the preprocessed frame; calculating a set of statistics for one or more connected components in the set of connected components, wherein calculating the set of statistics comprises gathering statistics of directions and values of gradients of intensity along a perimeter of a connected component of the one or more connected components; making a decision for the one or more connected components, using the calculated set of statistics, if the connected component is a light spot; and determining, using a processor, whether the frame comprises the glare in view of the decision for the one or more connected components.

2 . The method of claim 1 , wherein the preprocessing of the frame comprises binarizing the frame, wherein the frame is binarized based upon a threshold value.

3 . The method of claim 2 , wherein the threshold value is adjusted based on a previously analyzed frame of image data in an iterative manner.

4 . The method of claim 1 , wherein the calculating the set of statistics is performed at once for all of the connected components.

5 . The method of claim 1 , wherein the calculating the set of statistics is performed in an iterative manner until all glares are detected in the frame.

6 . The method of claim 1 , wherein the making the decision further comprises removing from further consideration all light spot connected components, wherein light spots connected components contain a recognizable signal.

7 . The method of claim 1 , wherein the set of statistics comprises one of a ratio of black pixels to a number of pixels of a perimeter of the connected component or an average intensity of the connected component.

8 . The method of claim 1 , further comprising:
sorting the set of connected components based upon a size of the components; and excluding connect components less than a predetermined size from the set of connected components.

9 . The method of claim 5 , wherein the calculating the set of statistics further comprising:
determining a second set of statistics for one or more of the connected components that remain in the set of connected components after removing from further consideration all light spot connected components based upon the decision making; and determining for each of the one or more of the connected components, using a trained classifier and the calculated second set of statistics, if the one or more of the connected components is a light spot.

10 . The method of claim 1 , wherein the frame of image data comprises a frame of a video sequence.

11 . The method of claim 1 , wherein the preprocessing of the frame comprises downsampling an analyzed frame.

12 . The method of claim 11 , wherein the downsampling is performed by sequentially applying a mask, or by subsequent smoothing and interpolation.

13 . The method of claim 1 , wherein the preprocessing of the frame comprises applying morphological filter, wherein the morphological filter increases a smoothing degree.

14 . The method of claim 1 , wherein the decision making comprises applying a trained classifier.

15 . The method of claim 1 , further comprising approximating the one or more connected components within an ellipsoid of inertia, wherein statistics for some properties of the ellipse and its smaller version is calculated.

16 . The method of claim 15 , wherein calculating the set of statistics from some properties of the ellipse and its smaller version comprises collecting and analyzing a number of supporting and rejecting brightness fluctuations on a circuit around the spot, statistics for a sample on a contour of an ellipse of scalar products of a brightness gradient and a unit vector from point to center, and statistics of pixels brightness along boundaries of the ellipses.

17 . The method of claim 1 , further comprising using for training a classifier for information about an amount of incorrect or uncertain recognized characters from OCR recognition.

18 . The method of claim 1 , further comprising suggesting for user a set of possible actions, wherein possible actions include instructions for changing one or more shooting conditions or settings to receive an image with better quality.

19 . The method of claim 1 , further comprising:
segmenting the frame into regions of image data; determining for a connected component a first set of regions of image data that contain this connected component; determining a third set of statistics based upon regions in the first set of regions for the connected component; determining a related statistic based upon one or more regions of the image data related to the first set of regions for the connected component; and comparing the third set of statistics to the related statistic determined therein.

20 . The method of claim 19 , wherein the third set of statistics is a signal-to-noise ratio of the regions in the first set of regions for the connected component, and wherein the related statistic is a signal-to-noise ratio of the one or more regions related to the regions in the first set of regions containing the connected component.

21 . The method of claim 20 , wherein the glare is determined to be present in the regions in the first set of regions for the connected component when the signal-to-noise ratio of the first set of regions for the connected component is more than a predetermined amount less than the signal-to-noise ratio of the one or more related regions of the image data.

22 . The method of claim 21 , further comprising determining the one or more regions of the image data by selecting regions surrounding the regions in the first set of regions.

23 . The method of claim 1 , further comprising:
determining an area of detected glare in the frame; altering the frame to visually represent the area of detected glare; displaying the altered frame; receiving input from a graphical user interface that changes the area of the glare; and updating parameters of a classifier to detect glare based upon the received input.

24 . The method of claim 23 , wherein the area of the detected glare is shown as a line representing a boundary of the detected glare.

25 . The method of claim 23 , wherein the area of the detected glare is enlarged based upon the received input.

26 . The method of claim 23 , wherein the area of the detected glare is reduced based upon the received input.

27 . The method of claim 1 , further comprising:
determining a message corresponding to detected glare to reduce or correct the detected glare; and displaying the message.

28 . The method of claim 1 , further comprising:
determining a level of defects within the frame based upon an analyzing of regions of image data to determine types of defects contained within the regions of image data; determining the level of defects within the frame is below a predetermined threshold; and saving the frame in long term memory based upon the determining the level of defects within the frame is below a predetermined threshold.

29 . The method of claim 1 , wherein detecting glare in a frame of image data is performed for a second preprocessed frame.

30 . A syst

Assignees

• Abbyy Dev Llc

Inventors

• Bocharov Konstantin
• Kostyukov Mikhail

Classifications

• G06V10/50

  by performing operations within image blocks; by using histograms, e.g. histogram of oriented gradients [HoG]; by summing image-intensity values; Projection analysis · CPC title

• G06V10/30

  Noise filtering · CPC title

• G06V10/431

  Frequency domain transformation; Autocorrelation · CPC title

• G06T5/005

  Physics · mapped topic

• G06T2207/20021

  Dividing image into blocks, subimages or windows · CPC title