System and methods for automated photosensitivity detection
The systems and methods described herein provide for an automated photosensitivity detection system (PDS) configured to automatically execute processes for flash detection and pattern detection of a video. PDS outputs an analysis result for each type of pattern detection analysis for the video. The PDS can execute each type of pattern detection analysis independently of the other pattern detection processes. Each pattern detection process is a distinct process that can be calculated without reference to the other processes. The final analysis result can aggregate the results of each detection process executed by the PDS.
1 . A system comprising:
one or more processors; and
one or more memory devices, wherein the one or more memory devices are communicatively coupled to the one or more processors, the one or more memory devices storing computer-executable instructions including at least an automated photosensitivity pattern detection module, wherein execution of the computer-executable instructions configure the one or more processors to:
receive video data comprising a plurality of video frames arranged sequentially, each individual frame comprising a plurality of pixels;
convert each frame of the video data from a first colorspace to a second colorspace;
for each frame within the video data,
detect a pattern region within the frame;
validate a pattern within the pattern region of the frame;
determine whether the pattern violates a pattern criteria;
in response to a determination that the pattern violates the pattern criteria within the frame, track a number of frames that include the pattern;
determine whether the number of frames satisfies at least one pattern threshold for the video data; and
output an indication of whether the video data satisfies the at least one pattern threshold.
2 . The system of claim 1 , wherein the pattern criteria defines a number of repeating elements, a pattern area threshold, and a threshold luminance difference in relative luminance between lighter and darker elements of the pattern.
3 . The system of claim 1 , wherein the at least one pattern threshold defines a threshold period of time, wherein execution of the computer-executable instructions further configure the one or more processors to determine whether the pattern is displayed for greater than the threshold period of time.
4 . The system of claim 1 , wherein to detect a pattern region within the frame, execution of the computer-executable instructions further configure the one or more processors to,
determine a Fourier transform of the frame;
determine a magnitude spectrum based on the Fourier transform;
determine a power spectrum based on the Fourier transform; and
detect a pattern region of the pattern based at least in part on the magnitude spectrum and the power spectrum.
5 . The system of claim 4 , wherein to detect a pattern region within the frame, execution of the computer-executable instructions further configure the one or more processors to,
determine an inverse Fourier transform of the frame to generate a modified image of the frame;
determine an absolute difference between the frame and the modified frame; and
detect the pattern region based at least in part on an absolute difference threshold applied to the difference between the frame and the modified frame.
6 . The system of claim 5 , wherein to detect a pattern region within the frame, execution of the computer-executable instructions further configure the one or more processors to,
convert the modified image into a binary image using an Otsu threshold;
apply at least one of dilation or erosion operations to the binary image resulting in an eroded image;
apply a contour operation on the eroded image; and
generate a pattern region mask based at least in part on the contour operation, wherein the pattern region mask defines boundaries of the pattern region within the frame.
7 . The system of claim 1 , wherein to validate a pattern within the pattern region of the frame, execution of the computer-executable instructions further configure the one or more processors to,
determine a number of dark and light components of the pattern;
determine an average luminance of the dark and light components of the pattern; and
determine a relative luminance between the dark and light components of the pattern.
8 . The system of claim 7 , wherein execution of the computer-executable instructions further configure the one or more processors to determine the number of dark and light components of the pattern using an Otsu threshold.
9 . The system of claim 1 , wherein execution of the computer-executable instructions further configure the one or more processors to determine a pattern area based on the pattern region and determine whether the pattern area satisfies a pattern area threshold.
10 . The system of claim 1 , wherein the first colorspace is RGB and the second colorspace is greyscale.
11 . A computer-implemented method to for photosensitivity flash detection comprising:
receiving video data comprising a plurality video frames arranged sequentially, each individual frame comprising a plurality of pixels;
converting each frame of the video data from a first colorspace to a second colorspace;
for each frame within the video data,
detecting a pattern region within the frame;
validating a pattern within the pattern region of the frame;
determining whether the pattern violates a pattern criteria;
in response to a determination that the pattern violates the pattern criteria within the frame, tracking a number of frames that include the pattern;
determining whether the number of frames satisfies at least one pattern threshold for the video data; and
outputting an indication of whether the video data satisfies the at least one pattern threshold.
12 . The method of claim 11 , wherein the pattern criteria defines a number of repeating elements, a pattern area threshold, and a threshold luminance difference in relative luminance between lighter and darker elements of the pattern.
13 . The method of claim 11 , wherein the at least one pattern threshold defines a threshold period of time, and further comprising determining whether the pattern is displayed for greater than the threshold period of time.
14 . The method of claim 11 , wherein detecting a pattern region within the frame, further comprises,
determining a Fourier transform of the frame;
determining a magnitude spectrum based on the Fourier transform;
determining a power spectrum based on the Fourier transform; and
detecting a pattern region of the pattern based at least in part on the magnitude spectrum and the power spectrum.
15 . The method of claim 14 , wherein detecting a pattern region within the frame, further comprises,
determining an inverse Fourier transform of the frame to generate a modified image of the frame;
determining an absolute difference between the frame and the modified frame; and
detecting the pattern region based at least in part on an absolute difference threshold applied to the difference between the frame and the modified frame.
16 . The method of claim 15 , wherein detecting a pattern region within the frame, further comprises,
converting the modified image into a binary image using an Otsu threshold;
applying at least one of dilation or erosion operations to the binary image resulting in an eroded image;
applying a contour operation on the eroded image; and
generating a pattern region mask based at least in part on the contour operation, wherein the pattern region mask defines boundaries of the pattern region within the frame.
17 . The method of claim 11 , wherein validating a pattern within the pattern region of the frame, further comprises:
determining a number of dark and light components of the pattern;
determining an average luminance of the dark and light components of the pattern; and
determining a relative luminance between the dark and light components of the pattern.
18 . The method of claim 17 further comprising determining the number of dark and light components of the pattern using an Otsu threshold.
19 . The method of claim 11 further comprising determining a pattern area based on the pattern region and determine whether the pattern area satisfies a pattern area threshold.
20 . A non-transitory computer readable medium storing instructions that, when executed by one or more processors, causes to perform a method comprising:
receiving video data comprising a plurality of video frames arranged sequentially, each individual frame comprising a plurality of pixels;
converting each frame of the video data from a first colorspace to a second colorspace;
for each frame within the video data,
detecting a pattern region within the frame;
validating a pattern within the pattern region of the frame;
determining whether the pattern violates a pattern criteria;
in response to a determination that the pattern violates the pattern criteria within the frame, tracking a number of frames that include the pattern;
determining whether the number of frames satisfies at least one pattern threshold for the video data; and
outputting an indication of whether the video data satisfies the at least one pattern threshold.