# US 12,181,603 B2 · Method and apparatus for high-confidence people classification, change detection, and nuisance alarm rejection based on shape classifier using 3D point cloud data > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 12,181,603 B2 - Document ID: US12181603B2 - Status: Granted - Kind: B2 - Application number: 17/206,577 - Filed: 2021-03-19 - Published: 2022-02-03 - Granted: 2024-12-31 - Earliest claimed: 2020-07-29 - Assignees: The Johns Hopkins University - Applicants: The Johns Hopkins University - Inventors: Bruce P. Mosier; Lee S. Stearns; Daniel D. Winker; Nicolas Norena Acosta - Web page: https://patentgrep.com/patent/US12181603B2 ## CPC - G01S 7/4802: {using analysis of echo signal for target characterisation; Target signature; Target cross-section} - G01S 17/42: Simultaneous measurement of distance and other co-ordinates (indirect measurement G01S17/46) - G01S 17/89: for mapping or imaging - G08B 29/24: Self-calibration, e.g. compensating for environmental drift or ageing of components - G08B 13/181: using active radiation detection systems - G08B 29/185: {Signal analysis techniques for reducing or preventing false alarms or for enhancing the reliability of the system} - G01S: RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES - G08B: SIGNALLING SYSTEMS, e.g. PERSONAL CALLING SYSTEMS; ORDER TELEGRAPHS; ALARM SYSTEMS ## Abstract A detector may include processing circuitry configured to receive monitoring data associated with an environment being monitored, where the monitoring data includes a series of frames of three-dimensional point cloud information. The processing circuitry may be further configured to perform image segmentation on the monitoring data to generate segmented three-dimensional data associated with the environment, identify one or more objects in the environment based on subtracting background from the segmented three-dimensional data, convert data associated with the one or more objects from the segmented three-dimensional data into two-dimensional object data, and perform object classification on the two-dimensional object data. ## Claims ### Claim 1 A detector comprising processing circuitry configured to: receive monitoring data associated with an environment being monitored, the monitoring data comprising a series of frames of three-dimensional point cloud information, perform image segmentation on the monitoring data to generate segmented three-dimensional data associated with the environment responsive to receiving the monitoring data; identify one or more objects in the environment based on subtracting background from the segmented three-dimensional data after generating the segmented three-dimensional data; convert data associated with the one or more objects from the segmented three-dimensional data into two-dimensional object data after the identification of one or more objects in the environment; and subsequently, perform object classification on the two-dimensional object data. ### Claim 2 (depends on 1) The detector of claim 1, wherein the monitoring data is received from a LIDAR sensor, and wherein the three-dimensional point cloud information is generated via LIDAR from the LIDAR sensor. ### Claim 3 A detector comprising processing circuitry configured to: receive monitoring data associated with an environment being monitored, the monitoring data comprising a series of frames of three-dimensional point cloud information, perform image segmentation on the monitoring data to generate segmented three-dimensional data associated with the environment; identify one or more objects in the environment based on subtracting background from the segmented three-dimensional data; convert data associated with the one or more objects from the segmented three-dimensional data into two-dimensional object data; and perform object classification on the two-dimensional object data, wherein identifying the one or more objects comprises comparing frames of the segmented three-dimensional data to subtract data associated with unchanged segments over a time averaged period and retain data associated with changed segments over the time averaged period. ### Claim 4 (depends on 1) The detector of claim 1, wherein converting data associated with the one or more objects from the segmented three-dimensional data into the two-dimensional object data comprises forming two-dimensional shapes from the one or more objects from the segmented three-dimensional data to define the two-dimensional object data. ### Claim 5 (depends on 4) The detector of claim 4, wherein the processing circuitry is further configured to classify the two-dimensional object data based on a size of the two-dimensional shape. ### Claim 6 (depends on 4) The detector of claim 4, wherein the processing circuitry is further configured to classify the two-dimensional object data based on a velocity of the two-dimensional shape. ### Claim 7 (depends on 1) The detector of claim 1, wherein performing image segmentation comprises performing nearest neighbor segmentation on each of the frames to identify candidate objects. ### Claim 8 (depends on 7) The detector of claim 7, wherein performing object classification comprises employing a neural network trained on a training data set to enable the candidate objects to be classified based on similarity to images in the training data set. ### Claim 9 (depends on 8) The detector of claim 8, wherein the processing circuitry is operably coupled to a synthetic shape generator configured to generate synthetic images of objects to at least partially define the training data set. ### Claim 10 (depends on 9) The detector of claim 9, wherein the training data set includes different perspective views of objects of interest and different levels of occlusion of the objects of interest. ### Claim 11 A method of detecting objects in an environment being monitored, the method comprising: receiving, from a sensor, monitoring data associated with the environment, the monitoring data comprising a series of frames of three-dimensional point cloud information, performing, by processing circuitry, image segmentation on the monitoring data to generate segmented three-dimensional data associated with the environment responsive to receiving the monitoring data; identifying, by the processing circuitry, one or more objects in the environment based on subtracting background from the segmented three-dimensional data after generating the segmented three-dimensional data; converting, by the processing circuitry, data associated with the one or more objects from the segmented three-dimensional data into two-dimensional object data after the identification of one or more objects in the environment; and subsequently performing, by the processing circuitry, object classification on the two-dimensional object data. ### Claim 12 (depends on 11) The method of claim 11, wherein the sensor is a LIDAR sensor, and wherein the three-dimensional point cloud information is generated via LIDAR from the LIDAR sensor. ### Claim 13 (depends on 11) The method of claim 11, wherein identifying the one or more objects comprises comparing frames of the segmented three-dimensional data to subtract data associated with unchanged segments over a time averaged period and retain data associated with changed segments over the time averaged period. ### Claim 14 (depends on 11) The method of claim 11, wherein converting data associated with the one or more objects from the segmented three-dimensional data into the two-dimensional object data comprises forming two-dimensional shapes from the one or more objects from the segmented three-dimensional data to define the two-dimensional object data. ### Claim 15 (depends on 14) The method of claim 14, wherein the processing circuitry is further configured to classify the two-dimensional object data based on a size of the two-dimensional shape. ### Claim 16 (depends on 14) The method of claim 14, wherein the processing circuitry is further configured to classify the two-dimensional object data based on a velocity of the two-dimensional shape. ### Claim 17 (depends on 11) The method of claim 11, wherein performing image segmentation comprises performing nearest neighbor segmentation on each of the frames to identify candidate objects. ### Claim 18 (depends on 17) The method of claim 17, wherein performing object classification comprises employing a neural network trained on a training data set to enable the candidate objects to be classified based on similarity to images in the training data set. ### Claim 19 (depends on 18) The method of claim 18, wherein the processing circuitry is operably coupled to a synthetic shape generator configured to generate synthetic images of objects to at least partially define the training data set. ### Claim 20 (depends on 19) The method of claim 19, wherein the training data set includes different perspective views of objects of interest and different levels of occlusion of the objects of interest.