Adaptive visual intelligence outdoor motion/occupancy and luminance detection system
View Patent ↗An image-based motion-occupancy sensor uses digital signal processing (DSP) techniques to perform algorithms on image data to enable an energy-delivery system, such as a lighting control system, to learn various environmental conditions that would conventionally result in false motion-occupancy detection. Motion of Interest (MOI) is determined by iteratively performing these algorithms and effectively filtering image data associated with non-MOI and making control decisions based only on relevant MOI.
1. A sensor fixed atop a structure having a light, the sensor comprising:
an image-capturing device that collects image data that includes data related to motion that is relevant to a control decision and data related to a wobbling event of the structure that is irrelevant to the control decision; and
a processing device that receives the image data, performs at least one algorithm on the image data to filter the data related to the wobbling event of the structure from the image data, and outputs an activation signal to the light based on the image data;
wherein the at least one algorithm includes determining a wobble state value, filtering data related to the wobbling event of the structure when the wobble state value is approximately equal to a first wobble state value, perform a first timeout for a first predetermined time period if the wobble state value is approximately equal a second wobble state value, and increment the first timeout to a second timeout for a second predetermined time period if the wobble state value is approximately equal the second wobble state value after the first timeout.
2. The sensor recited in claim 1 , wherein said processing device determines whether the sensor is moving based on the image data.
3. The sensor recited in claim 2 , wherein said processing device identifies image data corresponding to a relative motion of said sensor as the data related to motion that is irrelevant to the control decision.
4. The sensor recited in claim 1 , wherein said processing device determines whether an apparent motion detection is the result of an actual but irrelevant motion.
5. The sensor recited in claim 4 , wherein said actual but irrelevant motion corresponds to movement of a non-human.
6. The sensor recited in claim 4 , wherein said non-human includes one or more of an insect and a plant.
7. The sensor recited in claim 1 , wherein a processing time related to the algorithm is decreased as a result of said processing device learning at least one attribute of the image data.
8. A lighting system fixed atop a structure comprising:
a light fixture fixed atop the structure, the light fixture operable to provide illumination to an identified area;
a sensor module that collects and processes image data related to said identified area and determines a motion of interest in said identified area, the sensor module including a processing device that performs at least one algorithm on the image data to filter data related to motion that is relevant from data related to a wobbling event of the structure; and
a fixture control module that controls said light fixture based on said motion of interest determined by said sensor module;
wherein the at least one algorithm includes determining the wobble state value, filtering data related to the wobbling event of the structure when the wobble state value is approximately equal to a first wobble state value, perform a first timeout for a first predetermined time period if the wobble state value is approximately equal to a second wobble state value, and increment the first timeout to a second timeout for a second predetermined time period if the wobble state value is approximately equal to the second wobble state value after the first timeout.
9. The lighting system recited in claim 8 , wherein said sensor module further includes
an image-capturing device that collects image data that includes data corresponding to motion related to said motion of interest, and data corresponding to motion unrelated to said motion of interest.
10. The lighting system recited in claim 9 , wherein said motion unrelated to said motion of interest is apparent motion caused by movement of said sensor module.
11. The lighting system recited in claim 9 , wherein said motion unrelated to said motion of interest is motion of one or more of an insect, a bug and a plant.
12. The lighting system recited in claim 8 , wherein a processing time related to the algorithm is decreased as a result of said processing device learning at least one attribute of the image data.
13. The lighting system recited in claim 8 , wherein the structure is a light pole.
14. A method of controlling the operation of a lighting device fixed atop a structure, the method comprising:
capturing raw image data using a sensor module;
optimizing the raw image data for use in motion detection;
converting the optimized image data to gray scale;
performing, via a processing device, a Mixture of Gaussians (MOG) process on the converted image data to filter data related to motion that is relevant from data related to a wobble event of the lighting device, wherein the MOG process includes determining a wobble state value, filtering data related to the wobbling event of the structure when the wobble state value is approximately equal to a first wobble state value, perform a first timeout for a first predetermined time period if the wobble state value is approximately equal to a second wobble state value, and increment the first timeout to a second timeout for a second predetermined time period if the wobble state value is approximately equal to the second wobble state value after the first timeout;
determining if the lighting device and structure are wobbling based on the MOG process;
indicating whether a motion of interest has occurred based on a result of said determining; and
controlling said lighting device based on said indication of the motion of interest.
15. The method of claim 14 , wherein a processing time related to performing the MOG process is decreased as a result of said processing device learning at least one attribute of the image data.