WO2014009291A1

WO2014009291A1 - Vision based occupancy detection system and method

Info

Publication number: WO2014009291A1
Application number: PCT/EP2013/064312
Authority: WO
Inventors: Harshit Tiwari; Cristian Andreola
Original assignee: Osram GmbH
Current assignee: Osram GmbH
Priority date: 2012-07-12
Filing date: 2013-07-05
Publication date: 2014-01-16
Anticipated expiration: 2015-01-12

Description

VISION BASED OCCUPANCY DETECTION SYSTEM AND METHOD FIELD OF INVENTION

The present invention relates to a system and method of vi^¬ sion based occupancy detection using head-top circles. The inventive method has an occupancy detection algorithm for in^¬ door lighting control that detects humans in videos captured through a ceiling mounted camera utilizing a head-top circle detection-based approach. The method is scale and illumina^¬ tion invariant and works well even for top-view videos cap^¬ tured through a fish-eye lens-camera system. The human head is visible consistently from a ceiling mounted camera, and due to near spherical shape of human head, the projection of head in 2-D plain (e.g. the camera plane) is always near cir^¬ cular (disc) in shape. The present method utilizes this char^¬ acteristic of the human head top to accomplish occupancy de^¬ tection in a circle detection-based framework.

BACKGROUND ART

Various approaches for human detection techniques have been discussed in various literatures, ranging from non vision based detection techniques such as PIR motion sensors, acous- tic sensors, laser sensors to vision based detection tech^¬ niques such as camera based techniques.

As is known in the art, PIR motion sensors have been used to detect different human motion events. Heat signatures are captured for each of these events. A PIR motion sensor with ultra low power consumption has been known in the art. A motion sensing device has been designed to switch off lights when there is no motion. The hardware design has been opti- ^

mized to conserve power. A fusion of wireless nodes with PIR sensors has been proposed to track human motion. The computa^¬ tional and memory requirements are minimized in order to make the system amenable to running on resource constrained de- vices such as wireless nodes. The main drawback of PIR based sensing technology is that it cannot be used for counting the number of people in the scene. Another major drawback is that pose and fall detection of subjects is very tricky with simple PIR motion sensors. Also in case of very little or no mo- tion PIR motion sensors fail to detect human occupancy.

The acoustic systems designed for detection of acoustic events for describing human and social activity in meeting rooms. A fusion of audiovisual features has been proposed to improve the accuracy of acoustic event detection system. A human identification technique using acoustic micro-doppler signatures and band-pass sampling technique and gaussian mix^¬ ture model-based human identification technique was also pre^¬ sented by Z. Zhang and A. G. Andreou ("Human Identification Experiments Using Acoustic Micro-Doppler Signatures", The Ar^¬ gentine Conference on Micro-Nanoelectronics , Technology and Applications, Buenos Aires, Argentina, Sept, 2008.). The ac^¬ curacy of acoustics based human detection systems was not satisfactory enough for commercial deployment. It is also very difficult to analyze the pose/fall of a person and their unusual activities with acoustic system.

A method to detect human head-tops in videos is known using Kalman filter and mean-shift tracking. (H. Lu, R. Zhang and YW. Chen, "Head Detection and Tracking by Mean-shift and Kal^¬ man Filter", International Conference on Innovative Computing Information and Control (ICICIC 2008), pp. 357.) The technique employs the image color intensity information and the Local Binary Pattern (LBP) to construct a fourdimensional histogram representative of the color intensity values and the texture of the target under study. The new object location is then determined by Mean Shift iteration after the predict location is confirmed by Kalman filter. Color, texture, and motion features are integrated to track objects.

Another method for counting the people at crowded places us^¬ ing head-top detection using circle detection for top-view videos is presented in REMOTE HEAD COUNTING AND TRACKING IN

CROWDED SCENE VIA WWW/ INTERNET (Hsin Chia Fu, Jian-Rong Chen, Hsiao T. Pao Department of computer science, National Chiao- Tung University Hsinchu, Taiwan, 300 ROC) . Over-head images of crowded scene are collected from a set of head-down video cameras. Each collected over-head image is then analyzed by head-counting and tracking software for estimating the number of heads in the scene. If the number of heads exceeds a pre^¬ determined threshold, alarm signals and images of crowded scene are sent via Internet to the remote experts, so that the on-site situation can be examined and discussed to see if any further action may be needed.

In another literature a method based on a Bayesian fusion ap^¬ proach using laser range data and camera images is presented by Luciano Spinello and Roland Siegwart ("Human Detection Us^¬ ing Multimodal and Multidimensional Features", IEEE Interna^¬ tional Conference on Robotics and Automation, ICRA, pp. 3264- 3269, May 2008. ) . In addition many Patent literatures contains various methods for occupancy detection and dimensioning objects. Mechanical rulers are available in many stores, and they require contact to the surface that they measure. Optical methods are avail^¬ able for measuring various properties of a scene.

Various patents describe using optical triangulation to meas- ure the distance of objects from a video sensor. For example, in U.S. Pat. No. US5,255,064, multiple images from a video camera are used to apply triangulation to determine the dis^¬ tance of a moving target. In a PCT publication, WO2011151232, an optical system for occupancy sensing, and corresponding method is disclosed. The system includes a plurality of optical line sensors, each consisting of a linear array of light sensing elements; and an optical light integrating device that integrates light from rays with incidence angles subject to geometric con^¬ straints to be sensed by a light sensing element.

In an US publication, US2011267186, an optical method and system is disclosed. Therein successive images from an infra- red camera are analyzed to detect thermal characteristics of an occupant as well as movement.

In a PCT publication, WO2011091868, a system and method for 2D occupancy sensing is disclosed. The 2D occupancy system for determining a position of a user (e.g. a human, an object or an animal) in an environment according to embodiments of the invention includes a host device and a plurality of mo^¬ tion detection devices wherein the host device and the plu^¬ rality of motion detection devices are connected through a network. Each motion detection device has a viewing angle and the viewing angle of any motion detection device overlaps with the viewing angle of at least one other motion detection device . In another US publication US2004066500, occupancy detection and measurement, and obstacle detection using imaging tech^¬ nology is disclosed. Embodiments include determining occu^¬ pancy, or the presence of an object or person in a scene or space. If there is occupancy, the amount of occupancy is measured .

Most vision based algorithms rely on additional information for human detection such a motion cues or manually marked triggers/trip wires. In addition, most of these techniques use side view cameras for human detection in which many rich features of the human body such as the skin/face color (hue), shape and movement of the limbs etc. are available.

There is a need for cost effective and energy efficient method for human detection for various purposes especially for use in automation of lighting and HVAC control to save energy. Our technique aims to perform human detection in top- view videos using ceiling mounted cameras.

Ceiling mounted cameras offer a cheap alternative in terms of cost as well as computation complexity as it can be readily integrated to existing ceiling fixtures (e.g., luminaires, smoke detectors, air-conditioner vents, etc) .

The present method is fully automatic in the sense it does not require any additional manual inputs such as trip-wires or manual triggers for detection.

SUMMARY OF INVENTION

Energy efficiency of systems employing occupancy detection techniques depends on the accuracy of the human detection al^¬ gorithm. Design of such accurate algorithms is a non-trivial problem with vision based detection techniques and particu- larly challenging using ceiling mounted camera where only head-top is visible consistently. Herein disclosed is a novel vision-based occupancy detection algorithm using ceiling mounted cameras which utilizes the circularity of head shape for accurate detection. Experiments are conducted using an exhaustive video database and for varying illumination and ambience and showed that the said algorithm achieves a detec^¬ tion accuracy of up to 90%. In the present disclosure novel occupancy detection algorithm is introduced using ceiling mounted cameras. In such a sys^¬ tem, human detection implies detection of human head-tops since the only recognizable feature of human body available in all views is the head-top. Due to near-spherical shape of a human head, the projection of a head on a 2-D plane is al^¬ ways near-circular in shape. This characteristic of the human head-top is used to accomplish occupancy detection in a cir^¬ cle detectionbased framework. Described herein is a system for occupancy detection techniques exploiting the centre point's distribution for head- top circles in a region. The system includes atleast one ceiling mounted camera having fish-eye lens, a controller for executing an algorithm using ceiling mounted cameras that generates occupancy detection associated with the detection of human head tops predicted movement of occupants within each of the plurality of segments. The algorithm is a vision- based occupancy detection algorithm using gradient back propagation approach through ceiling mounted cameras which utilizes the circularity of head shape for accurate detec^¬ tion. The system further includes an output operably con^¬ nected to the controller to communicate the occupancy esti^¬ mates generated by the algorithm. In an exemplary embodiment of the present invention vision- based occupancy detection algorithm using gradient back propagation approach in conjunction with Centre Points Accu- mulation technique for detection of imperfect circles with specific application to head-top detection.

In another exemplary embodiment of the present invention vision-based occupancy detection algorithm using gradient back propagation approach in conjunction with homogeneity criterion for detection of imperfect homogenous circles with spe^¬ cific application to head-top detection.

In another aspect, described herein is a method of detecting occupancy in a region exploiting the centre points distribu^¬ tion for head-top circles in a region. The method includes modelling using occupancy detection algorithm using ceiling mounted cameras. Due to near-spherical shape of a human head, the projection of a head on a 2-D plane (e.g. the camera plane) is always near-circular in shape. The characteristic of the human head-top is used to accomplish occupancy detec^¬ tion in a circle detection-based framework. The method fur^¬ ther includes calculating modelbased predictions of occupancy within the region by detecting human head-tops by locating circles in grayscale images. The circles are detected by back propagating the gradients in the image and applying thresholds on accumulated points thereby detecting circle centres. The points are accumulated in a 3-D accumulator array of size (pl,ql,r3), where pi & ql are the width and the height of the image respectively and r3 is the number of radius values. The value of r3 depends upon the range of radius values used for circle detection. In another aspect, described herein is a system for detecting occupancy in a region using centre points distribution for head-top circles in a region. The system includes means for detecting each of the plurality of head top circles in a par- ticular region. The system further includes means for calculating model-based detections of occupancy within the region based on state equations that model occupancy of each region, detecting human head-tops by locating circles in grayscale images, further the circles are detected by back-propagating the gradients in the image and applying thresholds on accumu^¬ lated points thereby detecting circle centres. The points of the potential centre candidates are accumulated in a 3-D ac^¬ cumulator. After accumulation, a suitable threshold is applied to each 2-D (x,y) plane in accumulator array to detect circles of corresponding radii.

In another aspect, described herein is a computer readable storage medium encoded with a machine-readable computer pro^¬ gram code for generating occupancy detection for a region, the computer readable storage medium including instructions for causing a controller to implement the said method. The computer program includes instructions for calculating model- based detections of occupancy within the region based on state equations that model occupancy of the region, detecting human head-tops by locating circles in grayscale images, fur^¬ ther the circles are detected by back-propagating the gradi^¬ ents in the image and applying thresholds on accumulated points thereby detecting circle centres. The points of the potential centre candidates are accumulated in a 3-D accumu- lator. After accumulation, a suitable threshold is applied to each 2-D (x,y) plane in accumulator array to detect circles of corresponding radii. As it is insufficient to model the head-top as a unique cir^¬ cle due to variability of the human head in terms of its shape and size among different people, hence, the parametric definition of a circle is used in terms of centre coordinates and radius (x, y, r) to model the head-top circle as an im^¬ perfect circle with x, y and r as random variables. Circle centre coordinates obtained by gradient back propagation al^¬ gorithm form points in the (x, y, r) space. The circle centre points are accumulated in 3-D space to obtain the actual head-top circles. Further, the human head-tops appear as ho^¬ mogenous objects in top-view videos. The said property is used to distinguish head-top circles from random circles formed due to clutter in the scene. BRIEF DESCRIPTION OF THE ACCOMPANYING DRAWINGS

Embodiments of the invention are illustrated by way of exam^¬ ple, and not by way of limitation, in the figures of the ac^¬ companying drawings wherein -

Fig 1 illustrates an overview of the disclosed system. The Controller (1) connected to a ceiling mounted video camera (2) having fish-eye lens system. The Controller also controls the switches for lighting/HVAC according to occupancy detec- tion.

Fig 2 illustrates gradient back propagation for circular ob- j ects

Fig 3 illustrates circle detection using gradient back propa^¬ gation

Fig 4 illustrates distribution of centre candidates for dif^¬ ferent values of radius

Fig 5 illustrates a 3-D volume for concentrating centre candidates along (x, y, r) dimensions; Fig 6 illustrates homogeneity criterion gradient sums using circular integrator filters;

DETAILED DESCRIPTION

Advances in the fields of computer vision & image processing have led to generation of significant research interest in the area of vision based occupancy detection. Human occupancy detection is an integral component for various control appli- cations e.g. automatic lighting control, HVAC control etc.

Most of the lighting control systems rely on occupancy detec^¬ tion techniques to control lighting in a specified region.

Further energy-efficient lighting system design has for long been an active research area. Recently significant research interest has also been generated in the area of lighting con^¬ trol systems for outdoor as well as indoor lighting equip^¬ ment's. Most of the lighting control systems rely on occu^¬ pancy detection techniques to control lighting in a specified region. Herein disclosed is a novel occupancy detection algo^¬ rithm for indoor lighting control that detects humans in vid^¬ eos captured through a ceiling mounted camera utilizing a head-top detection-based approach. The disclosed method is scale and illumination invariant and works well even for top-view videos captured through a fish- eye lens-camera system. The method gives a detection accuracy of 90% (Lower of the two: Recall rate- 90% & Precision rate- 95%) . Experiments have been carried out for top-view videos of 15 people taken at two different locations, with varying illumination, ceiling height and ambience. The head-top de^¬ tection in top-view videos is performed utilizing a novel circle detection-based approach. The work specifically fo- cuses on a novel extension to the head-top detection algo^¬ rithm which improves the accuracy of the overall system by about 7%.

The term 'region' is used throughout the description to refer to both a region as well as various sub-divisions of the re^¬ gion. For instance, in the exemplary embodiment shown in figures, the term 'region' refers to both the area in general as well as to the individual sub-regions or zones e.g. rooms. Therefore, generating occupancy detection for the region would include generating occupancy estimates for each of the individual zones.

In addition, the term 'occupancy detection' is used throughout the description and refers generally to output related to occupancy. Therefore, an occupancy detection for a region may include data such as a number of occupants within the region, a probability associated with all possible occupancy levels associated with the region changes in occupancy, detecting something (human) ; catching sight of human, data indicative of the reliability of confidence associated with occupancy (e.g., covariance) , as well as other similarly useful data related to occupancy.

The application discloses a system and method for detecting occupancy based on exploiting the centre point's distribution for head-top circles. In an exemplary embodiment, occupancy is detected based solely on Gradient Back Propagation Ap^¬ proach. This application expands upon the scope of the prior art, disclosing additional embodiments and methods of imple^¬ menting occupant detection. Occupancy detection has traditionally been accomplished using either vision-based (e.g., using mast/ceiling mounted cameras or non vision-based techniques such as PIR, Ultrasonic, acoustic, etc. Vision-based occupancy detectors have been known to be cost effective and power efficient for both in^¬ door/outdoor scenarios, where a camera is mounted on a mast with a clear view of entry/exit points or locations with movement of people. For indoor scenarios, these sensors can be used to not only detect human occupation but they can also be useful for people counting, detection of unusual activity or possible fall of a person (e.g., in assisted living sce^¬ narios) . The disclosed novel occupancy detection algorithm using ceil^¬ ing mounted cameras in a system as shown in fig. 1. In such a system, human detection implies detection of human head-tops since the only recognizable feature of human body available in all views is the head-top. Due to near-spherical shape of a human head, the projection of a head on a 2-D plane (e.g. the camera plane) is always near-circular in shape. The char^¬ acteristic of the human head-top to accomplish occupancy de^¬ tection in a circle detection based framework is exploited. To this end, a comparative analysis of two techniques is pre- sented for circle detection in images namely, Hough circles and gradient back propagation. On analysis, it may be seen that the gradient back propagation algorithm yields 40% re^¬ duction in error-rate compared to Hough circlesbased approach for human detection problem.

In top-view videos there are no rich features available for human detection. The conventional features like the skin color (hue) , the face of the person, the shape and the move^¬ ment of the limbs, etc. cannot be used in this view. A lot of human detection techniques rely on facial features and limb shape/movements for detecting human presence. In the top- view, head-top is the only feature that is visible consis^¬ tently in all poses for human detection. Hence the problem of human detection boils down to human head-top detection in top-view. A human head resembles solid sphere in shape and the projection of a sphere is always circular at all angles. Head-top detection thus implies finding circles (disks) in top-view videos. We present an analysis of the Gradient back- propagation based approach for circle detection in the context of person detection.

In the discussion above, a circle detection algorithm for grayscale images utilizing the gradient direction alongwith the gradient magnitude is disclosed. The algorithm also pre^¬ sents a fixed computational complexity for a given image resolution . The Fig. 2 illustrates the gradient back propagation for cir^¬ cular objects. The disclosed algorithmic approach detects hu^¬ man head-tops by locating circles in grayscale images. The circles are detected by back-propagating the gradients in the image and applying thresholds on accumulated points thereby detecting circle centers. The points are accumulated in a 3-D accumulator array of size (pl,ql,r3), where pi & ql are the width and the height of the image respectively and r3 is the number of radius values. The value of r3 depends upon the range of radius values used for circle detection. After accu- mulation, a suitable threshold can be applied to each 2-D

(x,y) plane in accumulator array to detect circles of corre^¬ sponding radii. However, in order to apply the said approach for detection of head-top circles, which are imperfect cir^¬ cles, further improvements need to be done to the algorithm.

For head-top circles, the gradient back-propagation approach does not yield a unique center candidate. Instead, it returns a region around the actual circle center with all the points in the region satisfying the circle detection threshold. For perfect (near-perfect) circles, one can get clear local maxima in the accumulator array at the centre of the circle. Hence the other centre candidates can be suppressed in the region by finding local maxima around the centre in the accu^¬ mulator array. However, for imperfect circles (e.g. head-top circles) , the region around the centre in the accumulator ar^¬ ray typically consists of a distribution of points with high accumulation values. Hence, a simple approach to find local maxima may not fetch the exact centre coordinates. Similar distribution is observed for different values of radius close to the actual radius value. This distribution of accumulation values along the image axes (x,y) for different radii values for the original image in fig.l has been depicted in fig.4.

Fig. 4 illustrates the distribution of centre candidates for different values of radius. It is clear from fig. 5 that for imperfect circles, candidate center points are distributed in a 3-dimensional space constituting the 2 image axes (x,y) and a third radial axis. For a 2-dimensional distribution of cir^¬ cle points, one can concentrate the centre candidates using a mexican hat-shaped 2-D filter. This essentially finds the centroid of the distribution of the center candidates in the 2-D accumulator space. A threshold can be applied to the fil- tered accumulator array to find the actual center points. We can apply the same principle in 3-D space (x,y, radius) to find centroid of the distribution of the center candidates. In order to apply centroiding on center candidates in 3-D space one can use a 3-D filter as shown in fig 5 which illus- trates a 3-D volume for concentrating centre candidates along (x,y, radius) dimensions. As shown in fig. 5, an ellipsoid has been selected as the 3-D volume for integrating center candidates in the accumulator array, which is consistent with the choice of a circular fil^¬ ter along the 2-D image axes (x,y) . The span of the volume is longer along the radial axis as found through experiments that the distribution of center candidates is spread across a longer range on the radial axis compared to the image axes. The ellipsoid shaped 3-D filter is applied to accumulate the potential center candidates in the 3-D (x,y,r) accumulator array. The accumulation values at all points within the said ellipsoid in the neighborhood of a point in the accumulator array are integrated. A threshold is then applied to the in^¬ tegrated value at each pixel location to find actual circle centers. This technique yields about 7% improvement in the detection accuracy for head-top circles.

In the gradient back propagation approach the direction of the gradient vector is used, along with its magnitude, in circle detection framework. First the original RGB image is converted to grayscale and then filters are applied to find gradient vector at each point on the grayscale image. The gradient vector can be represented as (gr, θ) in the polar coordinate system, wherein gr is the magnitude of the gradi^¬ ent and & is the direction. For a dark colored circular ob- ject in the image, the gradient vectors at the object bound^¬ ary appear to emerge out of the object along radial lines as shown in fig . 2.

For such an object, if the gradient vectors are propagated backwards using a fixed distance, they all converge at the centre of the object if the distance value selected is equal to the radius of the circle. This is depicted in fig. 2(b) . This is the main concept behind circle detection using gradi- ent back propagation, the algorithm for which is explained below :

1) Convert the original RGB image to grayscale and apply two- dimensional gradient filters (sobel, canny etc.) to find gra^¬ dient vectors (r, Θ) at each point on the image.

2) Choose a range of radius values (rl, r2) to detect circles with radii lying in this range in the image. Here rl & r2 are integer values representing the minimum and maximum radius values respectively in terms of number of pixels. Hence the possible radius values for the chosen range are integers be^¬ tween (rl, r2) . Let r3 be the no. of radius values.

3) Initialize a 3-D accumulator array (accm) , of size

(pl,ql,r3) with zeros, where p and q are the width and the height of the grayscale image respectively.

4) For each value of radius (rv) in the chosen range (rl, r2), create a vector (rv, Θ +n) . This vector represents propagation of the gradient with distance rv in the reverse direction (Θ +n) i.e back-propagation of the gradient. Find this backpropagation vector (BPV) of the gradients for all points in the image.

5) Find projections of the BPV along the image axes i.e.

(x,y) coordinates and roundthem to integral pixel values. These can be calculated as: xl = [rv*cos (Θ +n) ] ; yl = [rv*sin(9 +n) ]

6) Add this BPV vector to the corresponding pixel location i.e. the point for which

the BPV vector was calculated. If (pl,ql) is the current pixel location, the resultant

location can be found as: p2 = pl+xl; q2 = ql+yl

7) Add the value of gradient magnitude at pixel location (pl,ql) in the grayscale image to the existing value at loca- tion (p2,q2,rv) in the accumulator array (accm) . Repeat the above steps for all pixels in the grayscale image and for all radius values (r3) .

8) The accumulator array gets populated with the gradient values of the original grayscale image. A suitable threshold can then be applied to each 2-D (x,y) plane in accumulator array to detect corresponding circles of radii between (rl, r2) . The entire process is depicted in fig. 3. The choice of threshold depends on the circularity of the object to be de- tected. For near perfect circles in the image, a high value of threshold is preferable, whereas, if the circles in the original image are imperfect or incomplete, it is better to choose a low value for this threshold. The head-top circles, as can be observed in the original im^¬ age fig. 2, are homogenous in nature i.e. the pixels in the head-top part of the image have similar intensity or gray^¬ scale values. The property of head-top circles is used to im^¬ prove the detection performance of our algorithm by applying a homogeneity constraint on the detected circles. This helps in eliminating false positive detection of circles formed with boundaries of heterogeneous objects or due to excessive clutter. There are a lot of different criteria to check the homogeneity of an object in an image, some of the simple ones being -variance/standard deviation of pixel values within the object, number of edge points within the object in the edge- map of the image, sum of gradient values within the object etc. Any one of these methods can be used depending upon the type of homogeneity to be checked. The sum of gradients as the homogeneity criteria is chosen as the gradients on the grayscale image for earlier steps has been already computed. After detecting circles in the image, gradients are inte- grated for all the points within the detected circles in the grayscale image. A suitable threshold is then applied to the gradient sums to find head- top circles among all the de^¬ tected circles. The gradient sums can also be calculated up^¬ front by applying circular 2-D filters on calculated gradient array for the grayscale image. This approach yields gradient sums for all the points in the image. The gradient sum values can be found at detected circle centres and apply threshold only on those selected values. It can be observed from fig.6 that homogenous regions of the grayscale image have lower values of gradient sums (dark re^¬ gions in the filtered image) compared to heterogene^¬ ous/cluttered regions. This has been used to distinguish ho^¬ mogenous head-top circles from false positive detections of heterogeneous circles. This is accomplished by selecting only those circles from all detected circle candidates which have a gradient sum value less than a predefined threshold. The value of the threshold for homogeneity depends on the circu^¬ larity of the object to be detected. For perfect (in shape) and homogenous circles (disks) , it is preferable to set a low value for the homogeneity threshold. For head-top circles, which are imperfect (but homogenous) circles, a higher value should be set as threshold. Therefore, as discussed above a novel head-top detection al^¬ gorithm is disclosed utilizing gradient back-propagation in grayscale images for use in conjunction with ceiling mounted camera (s) [at least one camera or a plurality of camera (s) operably connected] for occupancy detection applications. The algorithm performs person detection in top-view videos captured through ceiling mounted cameras having fisheye lens- camera system. Further the Centre Points Accumulation tech- nique is used for detection of imperfect circles has been proposed with specific application to head-top detection. This has been suggested as an extension to the said head-top detection algorithm utilizing gradient back-propagation. In addition, a homogeneity criterion for detection of imperfect homogenous circles is also disclosed with specific applica^¬ tion to head-top detection. This has been suggested as an ex^¬ tension to our head-top detection algorithm utilizing gradient back-propagation. The proposed technique detects head- tops in grayscale images for use in conjunction with ceiling mounted cameras for occupancy detection applications. In an^¬ other exemplary embodiment, the said algorithm and the pro^¬ posed techniques are used in conjunction for improving the output result. The algorithm performs person detection in top-view videos captured through ceiling mounted cameras hav- ing fisheye lens-camera system. This view has certain advan^¬ tages like occlusions are minimum and the coverage is good i.e. the number of cameras needed to monitor a certain area is least in this view. For the purposes of testing and records, a database of top- view videos has been generated by installing a ceiling mounted camera in a discussion room and a lab inside our of^¬ fice premises. The database captures variability in poses, head-shapes, ceiling height, and illumination conditions. The developed algorithm has been tested using the generated data^¬ base and a detection accuracy of 90% has been achieved. This method outperforms the Hough circles-based approach for cir^¬ cle detection as it utilizes more useful information about circular shapes as compared to the Hough-circles-based ap^¬ proach .

Although the foregoing description of the present invention has been shown and described with reference to particular em^¬ bodiments and applications thereof, it has been presented for purposes of illustration and description and is not intended to be exhaustive or to limit the invention to the particular embodiments and applications disclosed. It will be apparent to those having ordinary skill in the art that a number of changes, modifications, variations, or alterations to the in^¬ vention as described herein may be made, none of which depart from the spirit or scope of the present invention. The par^¬ ticular embodiments and applications were chosen and de- scribed to provide the best illustration of the principles of the invention and its practical application to thereby enable one of ordinary skill in the art to utilize the invention in various embodiments and with various modifications as are suited to the particular use contemplated. All such changes, modifications, variations, and alterations should therefore be seen as being within the scope of the present invention as determined by the appended claims when interpreted in accor^¬ dance with the breadth to which they are fairly, legally, and equitably entitled.

Claims

1. A system for occupancy detection exploiting the centre point's distribution for head top circles in a region, the system comprising:

at least one ceiling mounted camera having fish-eye lens operably connected to a controller configured for exe^¬ cuting an algorithm that generates occupancy detection associated with the detection of human head tops and predicted movement of occupants within each of the plu^¬ rality of regions, wherein the algorithm is a vision- based occupancy detection algorithm for detection of head-top circles utilizing gradient back propagation ap^¬ proach in conjunction with center point accumulation and/or homogeneity constraint; such that

an output operably connected to the said controller com^¬ municates the occupancy detection generated by the said algorithm.

2. A system for occupancy detection as claimed in claim 1, wherein the nearspherical shape of a human head top is used to accomplish occupancy detection in a circle de^¬ tection-based framework.

3. A system for occupancy detection as claimed in claim 1, wherein the image captured by the ceiling camera is con^¬ verted to grayscale and then gradient filters are ap^¬ plied to find gradient vector at each point on the gray^¬ scale image.

4. A system for occupancy detection as claimed in claim 1, wherein the gradient back propagation approach includes the direction of the gradient vector, along with its magnitude, in circle detection framework.

A system for occupancy detection as claimed in claim 3, wherein the gradient vector is represented as (gr, θ) in the polar coordinate system, wherein gr is the magnitude of the gradient and θ is the direction.

A system for occupancy detection as claimed in claim 3, wherein for an cicular object, the gradient vectors emerging out of the object along radial lines are propagated backwards using a fixed distance, they all converge at the centre of the object when the distance value selected is equal to the radius of the circle .

A system for occupancy detection as claimed in claim 1, wherein visionbased occupancy detection algorithm using gradient back propagation approach comprising the steps of :

converting the original RGB image to grayscale and apply two-dimensional gradient filters to find gradient vec^¬ tors (r, Θ) at each point on the image;

selecting a range of radius values (rl, r2) to detect circles with radii lying in this range in the image, where rl & r2 are integer values representing the minimum and maximum radius values respectively in terms of number of pixels;

initializing a 3-D accumulator array (accm) , of size (pl,ql,r3) with zeros, where p and q are the width and the height of the grayscale image respectively;

creating a vector (rv, Θ +n) for each value of radius (rv) in the selected range (rl, r2), where the said vec- tor represents propagation of the gradient with distance rv in the reverse direction (Θ +n) i.e back-propagation of the gradient;

finding back-propagation vector (BPV) of the gradients for all points in the image and the projections of the BPV along the image axes i.e. (x,y) coordinates and round them to integral pixel values;

adding the said BPV vector to the corresponding pixel location; adding the value of gradient magnitude at pixel location (pl,ql) in the grayscale image to the ex^¬ isting value at location (p2,q2,rv) in the accumulator array (acmm) ;

and

repeating the above steps for all pixels in the gray^¬ scale image and for all radius values (r3) .

A system for occupancy detection as claimed in claim 7, wherein the accumulator array (accm) gets populated with the gradient values of the original grayscale image.

A system for occupancy detection as claimed in claim 7, wherein the

projections of the BPV along the image are calculated as :

xl = [rv*cos (Θ +n) ] ; yl = [rv*sin(9 +n) ]

A system for occupancy detection as claimed in claim 7, wherein a suitable threshold is applied to each 2-D (x, y) plane in accumulator array to detect corresponding circles of radii between (rl, r2) and the choice of threshold depends on the circularity of the object to be detected .

A system for occupancy detection as claimed in claim 10, wherein for near perfect circles in the image, a high value of threshold is preferable, whereas, if the cir^¬ cles in the original image are imperfect or incomplete, a low value for threshold is selected.

A system for occupancy detection as claimed in claim 1, wherein the visionbased occupancy detection algorithm using gradient back propagation approach is used in conjunction with Centre Points Accumulation technique for detection of imperfect circles with specific application to head-top detection.

A system for occupancy detection as claimed in claim 1, wherein the visionbased occupancy detection algorithm using gradient back propagation approach is used in conjunction with homogeneity constraint for detection of imperfect homogenous circles with specific application to head-top detection.

A system for occupancy detection as claimed in claim 1, wherein the visionbased occupancy detection algorithm using gradient back propagation approach is used in conjunction with Centre Points Accumulation technique and homogeneity criterion.

A method of detecting occupancy in a region exploiting the centre points distribution for head-top circles in a region, the method includes steps of:

modelling using occupancy detection algorithm using at least one ceiling mounted camera;

calculating model-based predictions of occupancy within the region by detecting human head-tops by locating cir- cles in grayscale images wherein the circles are de^¬ tected by back propagating the gradients in the image and applying thresholds on accumulated points thereby detecting circle centres.

16. A method for occupancy detection as claimed in claim 15, wherein the nearsphercal shape of a human head top is used to accomplish occupancy detection in a circle de^¬ tection-based framework.

17. A method for occupancy detection as claimed in claim 15, wherein the gradient back propagation approach includes the direction of the gradient vector, along with its magnitude, in circle detection framework.

18. A method for occupancy detection as claimed in claim 15, wherein the image captured by the ceiling camera is con^¬ verted to grayscale and then filters are applied to find gradient vector at each point on the grayscale image.

19. A method for occupancy detection as claimed in claim 17, wherein the gradient vector is represented as (gr, Θ) in the polar coordinate system, wherein gr is the magnitude of the gradient and Θ is the direction.

20. A method for occupancy detection as claimed in claim 17, wherein for an object, the gradient vectors emerging out of the object along radial lines are propagated back^¬ wards using a fixed distance, converging at the centre of the object if the distance value selected is equal to the radius of the circle.

A method for occupancy detection as claimed in claim 15, wherein visionbased occupancy detection algorithm using gradient back propagation approach comprise the steps of:

creating a vector (rv, Θ +n) for each value of radius (rv) in the selected range (rl, r2), where the said vec^¬ tor represents propagation of the gradient with distance rv in the reverse direction (Θ +n) i.e back-propagation of the gradient;

adding the said BPV vector to the corresponding pixel location; adding the value of gradient magnitude at pixel location (pl,ql) in the grayscale image to the ex^¬ isting value at location (p2,q2,rv) in the accumulator array (accm) ;

and

A method for occupancy detection as claimed in claim 21, wherein the accumulator array (accm) gets populated with the gradient values of the original grayscale image.

A method for occupancy detection as claimed in claim 21, wherein the projections of the BPV along the image are calculated as:

xl = [rv*cos (Θ +n) ] ; yl = [rv*sin(9 +n) ]

A method for occupancy detection as claimed in claim 21, wherein the accumulation values at all points within the said cylinder in the neighbourhood of a point in the accumulator array are integrated and is given by eqn :

S = { (u,v,w) : TJ{X - uf + [y - vf I r-w I ≤th_r} 25. A method for occupancy detection as claimed in claim 21, wherein a suitable threshold is applied to each 2-D (x, y) plane in accumulator array to detect corresponding circles of radii between (rl, r2) and the choice of threshold depends on the circularity of the object to be detected.

26. A method for occupancy detection as claimed in claim 25, wherein for near perfect circles in the image, a high value of threshold is preferable, whereas, if the cir- cles in the original image are imperfect or incomplete, a low value for threshold is selected.

A method for occupancy detection as claimed in claim 15, wherein the visionbased occupancy detection algorithm using gradient back propagation approach is used in conjunction with Centre Points Accumulation technique for detection of imperfect circles with specific application to head-top detection.

A method for occupancy detection as claimed in claim 15, wherein the visionbased occupancy detection algorithm using gradient back propagation approach is used in conjunction with homogeneity criterion for detection of imperfect homogenous circles with specific application to head-top detection.

A method for occupancy detection as claimed in claim 15, wherein the visionbased occupancy detection algorithm using gradient back propagation approach is used in conjunction with Centre Points Accumulation technique and homogeneity criterion.

A system for detecting occupancy in a region using centre points distribution for head-top circles in a re^¬ gion, the system includes means for detecting each of the plurality of head top circles in a particular re^¬ gion;

means for calculating model-based detections of occu^¬ pancy within the region based on state equations that model occupancy of each region,

means for detecting human head-tops by locating circles in grayscale images, wherein the circles are detected by back-propagating the gradients in the image and applying thresholds on accumulated points thereby detecting cir- cle centres wherein the points of the potential centre candidates are accumulated in a 3-D accumulator and af^¬ ter accumulation, a suitable threshold is applied to each 2-D (x,y) plane in accumulator array to detect cir- cles of corresponding radii.

31. A computer readable storage medium encoded with a ma^¬ chine-readable computer program code for generating occupancy detection for a region, the computer readable storage medium including instructions for causing a controller to implement the method as claimed in claims 15 - 28.

The computer program product for occupancy detection exploiting the centre point's distribution for head top circles in a region, which includes instructions for calculating model-based detections of occupancy within the region based on state equations that model occupancy of the region;

detecting human head-tops by locating circles in grayscale images, further the circles are detected by back- propagating the gradients in the image and

applying thresholds on accumulated points thereby de^¬ tecting circle centres wherein the points of the poten^¬ tial centre candidates are accumulated in a 3- D accumu^¬ lator and after accumulation, a suitable threshold is applied to each 2-D (x,y) plane in accumulator array to detect circles of corresponding radii. 33. Use of a method according to any of claims 15 - 29 for modelling of cost effective and power efficient in^¬ door/outdoor scenarios and in lightning, HVAC control.