Above-water monitoring of swimming pools
An above-water system provides automatic alerting for possible drowning victims in swimming pools or the like. One or more electro-optical sensors are placed above the pool surface. Sequences of images are digitized and analyzed electronically to determine whether there are humans within the image, and whether such humans are moving in a manner that would suggest drowning. Effects due to glint, refraction, and variations in light, are offset automatically by the system. If a potential drowning incident is detected, the system produces an alarm sound, and/or a warning display, so that an operator can determine whether action must be taken.
1. A system for monitoring swimming pools for possible drowning victims, the system comprising:
a) a passive sensor, positioned above a surface of a pool, the sensor comprising means for receiving an image of an object located at or below the surface of the pool,
b) means for digitizing images received from the sensor,
c) a programmable computer connected to receive data from the digitizing means, the computer being programmed to analyze the images to determine whether the images indicate a presence of a drowning victim in the pool, and
d) means for alerting an operator of the presence of a drowning victim,
wherein the computer comprises entirely passive means for compensating for refraction of images caused by a roughened water surface due to surface waves in the pool, wherein the water surface defines a plurality of areas having different refraction properties,
wherein the computer comprises:
e) a spectral processor for decomposing each image into principal components,
f) a spatial processor which receives input from the spectral processor, the spatial processor comprising means for detecting predetermined shapes in the images, and
g) a temporal processor which receives input from the spatial processor, the temporal processor comprising means for analyzing sequences of images and for detecting the presence of predetermined movements of said predetermined shapes in said sequences of images,
wherein said predetermined shapes comprise human forms, and wherein said predetermined movements comprise vertical movements and vertical movements with rotation.
2. The system of claim 1 , wherein there are at least two sensors, and wherein the computer comprises means for comparing images received by the sensors so as to compute a distance to an object in said images.
3. The system of claim 1 , wherein the computer comprises means for enhancing quality of images by extracting principal components of the images, wherein said principal components are non-correlated, and analyzing at least some of said principal components to derive information about the images.
4. A method for monitoring swimming pools for possible drowning victims, the method comprising:
a) forming a sequence of digital images of a pool, each image being formed by a passive sensor positioned above a surface of the pool, the sensor comprising means for receiving an image of an object located at or below the surface of the pool,
b) analyzing said images to determine whether the mages indicate a presence of a drowning victim in the pool, and
C) alerting an operator of the presence of a drowning victim,
wherein step (b) includes passively compensating for refraction of images caused by a roughened water surface due to surface waves in the pool, wherein the water surface defines a plurality of areas having different refraction properties,
wherein step (b) includes:
d) decomposing each image into principal components,
e) analyzing each image component produced in step (d) to detect predetermined shapes in said image component, and
f) analyzing sequences of image components obtained from step (e) to detect predetermined movements of said predetermined shapes in said sequences of image components,
wherein said predetermined shapes are selected to the human forms, and wherein said predetermined movements are selected form the group consisting of vertical movements and vertical movements with rotation.
5. The method of claim 4 , wherein there are at least two sensors, and wherein step (b) includes comparing images received by the sensors so as to compute a distance to an object in said image.
6. The method of claim 4 , wherein step (b) includes enhancing quality of images by extracting principal components of the images, wherein said principal components are non-correlated, and analyzing at least some of said principal components to derive information about the images.
7. A method of monitoring a pool so as to detect possible drowning victims therein, the method comprising:
a) forming a sequence of digital images of a pool, each image being formed by a passive sensor positioned above a surface of the pool, the sensor comprising means for receiving an image of an object located at or below the surface of the pool,
b) decomposing each image into principal components so as to enhance quality of said images,
c) analyzing at least some principal components obtained from step (b) to detect a presence of predetermined shapes in said image,
d) analyzing sequences of images having been found to be likely to contain said predetermined shapes, to detect predetermined movements associated with drowning, wherein the analyzing steps include passively compensating for refraction of images caused by a roughened water surface due to surface waves in the pool, wherein the water surface defines a plurality of areas having different refraction properties, and
e) alerting an operator of the existence of a possible drowning event when said predetermined movements are detected.