Discontinuity detection in video data

A method of motion detection, comprising:

• receiving, from a camera, a first frame of video data representing a physical environment;
• receiving, from the camera, a second frame of the video data, wherein the second frame is consecutive to the first frame;
• dividing the first frame into a first plurality of blocks of pixels, the first plurality of blocks of pixels including a first block of pixels and a second block of pixels;
• dividing the second frame into a second plurality of blocks of pixels, the second plurality of blocks of pixels including a third block of pixels and a fourth block of pixels;
• determining a set of matching block pairs by matching each block of pixels of the first plurality of blocks of pixels with a corresponding block of pixels of the second plurality of blocks of pixels, wherein each matching block pair of the set of matching block pairs represents a same region of the physical environment;
• for each matching block pair of the set of matching block pairs, determining a sum of squared differences (SSD) of the matching block pair to generate SSD values of the set of matching block pairs;
• determining a first mean SSD value for the set of matching block pairs;
• determining a standard deviation of the SSD values of the set of matching block pairs;
• determining a first convolution of the first mean SSD value with a kernel K, wherein K=[(α−1)/2, α, (α−1)/2], where 0<α<1, wherein α is a peak significance factor used to weight the first convolution;
• determining a second convolution of the standard deviation of the SSD values with the kernel K;
• determining that a first result of the first convolution is greater than a mean threshold value, wherein the mean threshold value indicates an amount of change between two or more frames of image data that is not due to motion in the physical environment;
• determining that a second result of the second convolution is greater than a standard deviation convolution threshold value, wherein the standard deviation convolution threshold value indicates an amount of deviation between changes in the two or more frames of image data that is not due to motion in the physical environment;
• and
• determining that differences in the video data between the first frame and the second frame are due to an illumination change in the physical environment and not due to motion in the physical environment.