FORM 2
THE PATENTS ACT, 1970
(39 of 1970)
&
THE PATENT RULES, 2003
COMPLETE SPECIFICATION
(See Section 10 and Rule 13)
Title of invention:
SYSTEMS AND METHODS FOR AUTOMATICALLY DETECTING CAP INSULATORS
Applicant:
Tata Consultancy Services Limited
A company Incorporated in India under the Companies Act, 1956
Having address:
Nirmal Building, 9th floor,
Nariman point, Mumbai 400021,
Maharashtra, India
The following specification particularly describes the embodiments and the manner in which it is to be performed. 2

TECHNICAL FIELD
[0001] The embodiments herein generally relate to image analysis and more particularly to methods and systems for detecting cap insulators based on elliptical shapes identified from images.

BACKGROUND
[0002] Continuous monitoring leading to early identification and repair of damaged components of electrical power distribution systems are critical for uninterrupted power supply. Early detection of damage can prevent more severe breakdown in future and save time and huge cost of repair work of higher magnitude. Ceramic cap insulators used in high power transmission towers are suspended at great heights on pylons, and hence it is difficult to inspect them from the ground. Using Unmanned Aerial Vehicles (UAVs) for close inspection of insulators is both time and cost effective. But since videos obtained by UAV inspections can be long, an automated method to identify insulators and damage thereof can reduce huge manual effort.

SUMMARY
[0003] Embodiments of the present disclosure present technological improvements as solutions to one or more of the above-mentioned technical problems recognized by the inventors in conventional systems.

[0004] The present disclosure provides systems and methods for automatically detecting and identifying damages to ceramic cap insulators, from say outdoor aerial surveillance videos. The method is applicable to both strain and suspension types of insulator; is invariant to insulator orientation, size, number of caps and time of day and can be used for automatic labeling of insulators for further machine learning techniques.

[0005] In an aspect, there is provided a method comprising detecting edge segments representing curves, from at least one image of interest; and identifying elliptical shapes from pairs of curves that satisfy one or more pre-defined pairing conditions that result in an ellipse, the pairs of curves being identified from the detected edge segments.

[0006] In another aspect, there is provided a system comprising: one or more data storage devices operatively coupled to the one or more processors and configured to store
3

instructions configured for execution by the one or more processors to: detect edge segments representing curves, from at least one image of interest; and identify elliptical shapes from pairs of curves that satisfy one or more pre-defined pairing conditions that result in an ellipse, the pairs of curves being identified from the detected edge segments.

[0007] In yet another aspect, there is provided a computer program product comprising a non-transitory computer readable medium having a computer readable program embodied therein, wherein the computer readable program, when executed on a computing device, causes the computing device to: detect edge segments representing curves, from at least one image of interest; and identify elliptical shapes from pairs of curves that satisfy one or more pre-defined pairing conditions that result in an ellipse, the pairs of curves being identified from the detected edge segments.

[0008] In an embodiment of the present disclosure, the one or more pre-defined pairing conditions comprise: each edge segment in the pairs of curves is an approximated set of at least three or more line segments to ensure at least a degree of curvature; the pair of curves, when extrapolated towards each other converge to result in a convex region; and concave portions of the pairs of curves face each other.

[0009] In an embodiment of the present disclosure, identifying elliptical shapes is based on least squares optimization method for ellipse fitting wherein input to the least squares optimization method comprises the pairs of curves and output thereof comprises parameters for identifying elliptical shapes.

[0010] In an embodiment of the present disclosure, the method described herein above further comprises filtering the elliptical shapes to obtain candidate discs of cap insulators from the at least one image of interest based on pattern selection for linearly oriented shapes.

[0011] In an embodiment of the present disclosure, filtering the elliptical shapes is based on: the elliptical shapes having same size; orientation of the elliptical shapes; and centers of the elliptical shapes being aligned along a line perpendicular to the orientation.

[0012] In an embodiment of the present disclosure, the method described herein
4

above further comprises validating the filtered elliptical shapes as the candidate discs of cap insulators based on region growing method.

[0013] In an embodiment of the present disclosure, the method described herein above further comprises detecting damaged candidate discs of cap insulators based on inter-center distances between the centers of successive elliptical shapes.

[0014] It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the embodiments of the present disclosure, as claimed.

BRIEF DESCRIPTION OF THE DRAWINGS
[0015] The embodiments herein will be better understood from the following detailed description with reference to the drawings, in which:

[0016] FIG.1 illustrates an exemplary block diagram of a system for automatically detecting cap insulators in accordance with an embodiment of the present disclosure;

[0017] FIG.2 illustrates an exemplary flow diagram of a method for automatically detecting cap insulators in accordance with an embodiment of the present disclosure;

[0018] FIG.3 illustrates an exemplary representation of an output of pre-processing edge segments of elliptical shapes detected from images of interest in accordance with an embodiment of the method of the present disclosure;

[0019] FIG.4 illustrates an exemplary representation of elliptical shapes identified using a ellipse fitting method known in the art;

[0020] FIG.5A and FIG.5B illustrate exemplary representations of curve pairing and ellipse detection respectively in accordance with an embodiment of the present disclosure; and

[0021] FIG.6 illustrates detection of a damaged disc in accordance with an embodiment of the present disclosure.
5

[0022] It should be appreciated by those skilled in the art that any block diagram herein represent conceptual views of illustrative systems embodying the principles of the present subject matter. Similarly, it will be appreciated that any flow charts, flow diagrams, state transition diagrams, pseudo code, and the like represent various processes which may be substantially represented in computer readable medium and so executed by a computing device or processor, whether or not such computing device or processor is explicitly shown.

DETAILED DESCRIPTION
[0023] The embodiments herein and the various features and advantageous details thereof are explained more fully with reference to the non-limiting embodiments that are illustrated in the accompanying drawings and detailed in the following description. The examples used herein are intended merely to facilitate an understanding of ways in which the embodiments herein may be practiced and to further enable those of skill in the art to practice the embodiments herein. Accordingly, the examples should not be construed as limiting the scope of the embodiments herein.

[0024] The words "comprising," "having," "containing," and "including," and other forms thereof, are intended to be equivalent in meaning and be open ended in that an item or items following any one of these words is not meant to be an exhaustive listing of such item or items, or meant to be limited to only the listed item or items.

[0025] It must also be noted that as used herein and in the appended claims, the singular forms "a," "an," and "the" include plural references unless the context clearly dictates otherwise. Although any systems and methods similar or equivalent to those described herein can be used in the practice or testing of embodiments of the present disclosure, the preferred, systems and methods are now described.

[0026] Some embodiments of this disclosure, illustrating all its features, will now be discussed in detail. The disclosed embodiments are merely exemplary of the disclosure, which may be embodied in various forms.

[0027] Before setting forth the detailed explanation, it is noted that all of the discussion below, regardless of the particular implementation being described, is exemplary in nature, rather than limiting.
6

[0028] The present disclosure provides systems and methods that enable automatic detection of cap insulators and damage thereof. Porcelain discs of insulators are generally brown in color and are made from aluminum silicate mixed with plastic kaolin, feldspar and quartz to obtain the final hard composite and a glazed finish. Glazing is done to repel water, but it also reflects light causing a glowing effect. Also, local illumination variance, time of the day, and weather renders color an inconsistent feature across images. Depending on whether the insulator is of suspension or strain type, its orientation varies. Depending on the view, due to perspective projection in the 2D space, the discs appear circular or elliptical with varying major to minor axis ratio. They also may or may not overlap with each other, resulting in self-occlusion. The only definitive and consistent feature is that the projections of all the discs, closely approximated by ellipses, are of the same size, same orientation and align along a line perpendicular to orientation of the discs in a given image. The shape of the insulator, from any camera view, is thus naturally represented as a composite structure of various ellipses that meet the above criteria. In an aerial view, the background of an insulator typically comprises of transmission tower truss, which are linear structures with no elliptical shape within. Further, it may comprise of vegetation on earth, which though might have ellipse like shapes embedded within, the sizes and orientation of such ellipses is expectedly random. Hence the insulator can be easily isolated, with very minimal chance of pattern overlaps, from the background in any complex aerial scene. Also, importantly, the breakages of the discs can be identified easily as a deviation from this pattern.

[0029] Referring now to the drawings, and more particularly to FIGS. 1 through 6, where similar reference characters denote corresponding features consistently throughout the figures, there are shown preferred embodiments and these embodiments are described in the context of the following exemplary system and method.

[0030] FIG.1 illustrates an exemplary block diagram of a system 100 for automatically detecting cap insulators in accordance with an embodiment of the present disclosure. In an embodiment, the system 100 includes one or more processors 104, communication interface device(s) or input/output (I/O) interface(s) 106, and one or more data storage devices or memory 102 operatively coupled to the one or more processors 104. The one or more processors 104 that are hardware processors can be implemented as one or more microprocessors, microcomputers, microcontrollers, digital signal processors, central processing units, state machines, logic circuitries, and/or any devices that manipulate signals
7

based on operational instructions. Among other capabilities, the processor(s) is configured to fetch and execute computer-readable instructions stored in the memory. In an embodiment, the system 100 can be implemented in a variety of portable computing systems, such as laptop computers, notebooks, hand-held devices including cell phones, workstations, mainframe computers, servers, a network cloud and the like.

[0031] The I/O interface device(s) 106 can include a variety of software and hardware interfaces, for example, a web interface, a graphical user interface, and the like and can facilitate multiple communications within a wide variety of networks N/W and protocol types, including wired networks, for example, LAN, cable, etc., and wireless networks, such as WLAN, cellular, or satellite. In an embodiment, the I/O interface device(s) can include one or more ports for connecting a number of devices to one another or to another server.

[0032] The memory 102 may include any computer-readable medium known in the art including, for example, volatile memory, such as static random access memory (SRAM) and dynamic random access memory (DRAM), and/or non-volatile memory, such as read only memory (ROM), erasable programmable ROM, flash memories, hard disks, optical disks, and magnetic tapes. In an embodiment, one or more modules (not shown) of the system 100 can be stored in the memory 102.

[0033] FIG.2 illustrates an exemplary flow diagram of a method 200 for automatically detecting cap insulators in accordance with an embodiment of the present disclosure. In an embodiment, the system 100 comprises one or more data storage devices or memory 102 operatively coupled to the one or more processors 104 and is configured to store instructions configured for execution of steps of the method 200 by the one or more processors 104. In an embodiment, at step 202, the system 100 is configured to receive an input RGB (additive primary colors red green blue) frame.

[0034] In the absence of a priori information about the orientation and the size of insulator discs, it is imperative to identify all elliptical shapes in the image or RGB frame received in the step 202 and then narrow down on candidate discs of cap insulators based on their properties. In an embodiment, ellipse fitting (ElliFit) technique proposed by Prasad et al. (2013), a non-iterative least squares optimization method that works on partial curves having standard computational complexity of O(N) can be used for identifying elliptical shapes.
8

However, ElliFit technique limits itself to take as input, a mere set of edge points, into which it tries to fit an ellipse in least-squares manner. Hence, preprocessing has to be done on any aerial image, before a candidate set of edge segments can be passed on for fitting ellipses using ElliFit. These steps include RGB to gray scale conversion, histogram equalization, range filtering, isolation of continuous edges, and then splitting of edges into smaller fragments (referred as fragmentation). The last preprocessing step is performed so that edges belonging to different occluding objects, which are merged as a single edge segment, are split into groups belonging to different objects, as far as possible. Such an edge segment will not be a simple edge and hence not fit into any simple shape. When only edge segment detection is applied on an insulator, a continuous edge encompassing the whole insulator structure is obtained. The fragmentation in this case is necessary to separate edges of two or more overlapping ellipses, or an ellipse and a non-elliptical neighbor (metal links between the discs or any background object occluding a part of the insulator).

[0035] Step 204 of the method of the present disclosure involves edge segment extraction and pre-processing of extracted edge segments. Given the span of edges along the contour of any insulator, in an embodiment, a modified subset of preprocessing steps in accordance with the present disclosure, as summarized below is performed.

Step a: The image or RGB frame received at step 202 is converted to gray scale.
Step b: Canny edge detector with pre-determined lower and upper thresholds, say 0.1 and 0.2 respectively, is used to extract the edges in the image.
Step c: Kovesi’s method is used to obtain connected edge contours. A threshold is applied on the edge length to eliminate all edges smaller than say 10 pixels.
Step d: The contours are then approximated as a set of line segments using Ramer–Douglas–Peucker algorithm (RDP)-mod method.
Step e: Edge contours are further corrected by splitting edge curves at sharp turns and inflexion points. In an embodiment, step e is a two stage edge-splitting method. In the first stage, the edge is split when there is a sharp turn of 90° or more. Even after this, corners at some points, where the discs overlap, might still remain intact. So, inflexion points, which trace the change in the sign of the angle, are further used to separate overlapping discs. FIG.3 illustrates an exemplary representation of an output of step 204 of pre-processing edge segments 302 representing curves detected from images of interest in accordance with an embodiment of the method of the present disclosure. 9

[0036] FIG.4 illustrates an exemplary representation of elliptical shapes identified using an ellipse fitting method known in the art. As seen from FIG.4, if a set of points 402 is provided as an input to an ellipse fitting method such as ElliFit, possible ellipses (404, 406) that minimized the least square conditions are detected, which may not necessarily be the original ellipse to which the set of points 402 belong. This is due to digitization noise in the set of edge points that form the input curve. To overcome this impediment, at step 206, elliptical shapes are identified from pairs of curves that satisfy one or more pre-defined pairing conditions that result in an ellipse, the pairs of curves being identified from the detected edge segments.

[0037] In an embodiment, the one or more pre-defined pairing conditions can include:

Condition a: Each edge segment in the pairs of curves is an approximated set of at least three or more line segments to ensure at least a degree of curvature. This condition ensures that the edge segment has some degree of curvature and increases the probability of the edge segment being a part of an ellipse. In the penultimate step of preprocessing (Step d), every edge is approximated into a set of line segments. This facilitates calculation of the degree of curvature along the edge.
Condition b: The pair of curves, when extrapolated towards each other converge to result in a convex region. In other words, a second curve should lie within the search region of a first curve. If tangents are drawn at the ends of the first curve towards the concave side, the two end points of the second curve should lie within the region between two tangents. This ensures that the pair of curves, when extrapolated towards each other, do converge and give rise to a convex region such as an ellipse.
Condition c: concave portions of the pairs of curves face each other. This ensures that least squares minimization will converge to a lesser-error configuration, more likely to be true positive.
[0038] In an embodiment, at step 206, for ellipse fitting, input to the least squares optimization method comprises the pairs of curves satisfying one or more pairing conditions as described herein above and output comprises parameters for identifying elliptical shapes. In an embodiment, if the input to the ellipse fitting method has an optimal least square fit, the parameters obtained can include:

Length of the major axis (a) and length of the minor axis (b), wherein a ≧ b;
Orientation of the ellipse α, wherein α ∈ {– Π/2, Π/2} 10

Center of the ellipse (xc, yc), wherein the center should lie within the image.
Thus at step 206, all possible ellipses obtained from all possible pairs of curves subject to the pairing conditions that result in an ellipse are identified. FIG.5A and FIG.5B illustrate exemplary representations of pairing of curves 502 and detecting ellipse 504 respectively in accordance with an embodiment of the present disclosure. Multiple instances of the same ellipse can be formed by pairing of different fragments of the same ellipse.
[0039] In an embodiment, at step 208, the obtained possible elliptical shapes are filtered to obtain candidate discs of cap insulators based on pattern selection for linearly oriented shapes. Since cap insulators are characterized by ‘stringed discs’, the ellipses forming any suspension or strain type insulators would have discs/ellipses having the same size; the ellipses have same orientation since they are stringed and their centers are positioned along a line perpendicular to its orientation. Hence, to detect ‘stringed discs’, at step 208, the elliptical shapes are filtered based on the elliptical shapes having same size; orientation of the elliptical shapes; and centers of the elliptical shapes being aligned along a line perpendicular to the orientation by a step by step elimination process. Given the three features, size, orientation and center, the order of elimination is chosen such that efficiency of the method is maximized in terms of no. of iterations.

[0040] Since orientation is a single parameter α, in an embodiment, orientation may be the first feature to be considered in the elimination process. If the orientation is fixed, one parameter in the alignment of ellipse centers may be reduced as the ellipses have to be aligned perpendicular to the orientation. After these two steps of elimination, the size feature may be considered in the end. This particular order ensures that the search space of all detected ellipses is pruned fastest, to detect a linear constellation.

[0041] The background of images received at step 202 is generally transmission tower truss, occasional vegetation and sky. The ellipses formed by the contour of any background objects will be random. So the assumption that maximum number of ellipses that satisfy all the above conditions will belong to the insulator is most practical. Accordingly, ellipses are first grouped according to their orientation. Then, iteratively, for each group, the ellipses are filtered, the centers of which are aligned along a line perpendicular to the orientation. There may be many lines parallel to each other that are perpendicular to the ellipses’ orientation. A line that has maximum number of ellipses is chosen. Thirdly, from
11

among these ellipses a final group of ellipses that has maximum number of ellipses of the same size after eliminating overlapping ellipses is chosen. The steps of filtering ellipses and identifying final group of ellipses are repeated for every ellipse orientation. In accordance with the present disclosure, the group that has the maximum number of ellipses satisfying all the conditions will emerge as the best configuration of desired shape, and is hereafter referred to as the candidate group.

[0042] The color of pylon truss is significantly lighter when compared to the color of porcelain discs, in sunlight. In such a case, the detected edges of the discs, even when the truss edges intersect the disc region, will be prominent. However, self-shadowing (front view, concave part being in shadow) of truss may cause both the truss and the disc to have similar dark shades. This property is true even for any occluding vegetation in the background. The edges of the insulator discs will be weak in these cases. Hence it will lead to missed detection for specific discs. Thus, in accordance with the present disclosure shape alone is not the single discriminative feature to detect ellipses. In an embodiment, at step 210, the filtered elliptical shapes from step 208 are validated as the candidate discs of cap insulators by considering lightweight appearance features and performing region growing method.

[0043] In accordance with the present disclosure, specifically, if at least one candidate disc has been identified, neighborhood regions along its line of alignment, can be explored and matched for appearance. In an embodiment, this is done by matching a histogram pattern of identified disc with candidate discs in the neighborhood. This process uncovers those discs that are present but miss detection by consideration of shape alone. For matching the histogram pattern, in an embodiment, the initial step is to calculate an average distance dmean between the centers of all ellipses in the candidate group of ellipses. Every ellipse ec in the candidate group is taken and its histogram is calculated. Then a search is done for a matching ellipse in a touching, but not intersecting, parallelogram region along the line. The parallelogram region is of length (dmean – b/2) where b is the length of the minor axis of the ellipse in consideration, along the line of alignment. In an embodiment, the width of the parallelogram, along x or y axis is 11 pixels: 5 pixels on either side, perpendicular to the line of alignment. The search is performed along the front of the ellipse and back. Thus there are two search regions per ellipse. The three channel histogram, hc of the ellipse considered ec, and the three channel histogram hs of an elliptical region (same size and orientation as ec) in the search region, are compared by calculating the well-known appearance similarity
12

measure, Bhattacharyya distance dBhat, between the two. If an elliptical region has a Bhattacharyya distance less than 0.15 and it is the minimum in the search region, then this region is taken as a candidate and is added to the candidate group of ellipses, for next round of iterative growing.

[0044] To iteratively search and add, the region in the front and back of every ellipse in the candidate group is searched, including the added new candidate ellipses. To avoid replication, a candidate is eliminated if it is placed too close to any of the ellipses in the candidate group. In the present disclosure, this step facilitates recovery of ellipses that missed detection in the earlier steps.

[0045] As mentioned earlier, typical materials used in manufacturing insulators have very good insulation properties. However, they also have poor fracture strength making them the most damage-prone component of a power grid. The breakage in most cases is not a hole in the interior region of insulator, but a “chipping” effect into the interior, starting from the contour of the insulator. Such chipping / breakage can be visualized as some kind of partial occlusion. When a disc is positioned such that a major portion of its damage is captured by the camera, then very little edge length of the original unbroken ‘ellipse’ is captured by the camera. This is because there is further self-occlusion between discs of an insulator. Hence much of the un-occluded part of disc contour is a random-shaped edge arising out of breakage. This leads to missing of ellipse detection for this disc in at step 206 in most cases. The method of the present disclosure uses this ‘missing’ property to predict and locate damage in a disc.

[0046] In accordance with the present disclosure, pattern completion in the ellipses of the candidate group is examined for predicting and locating damaged/broken discs. Given any subsequence of natural numbers S ∈ N, if successive difference of adjacent numbers is considered in the subsequence over multiple iterations and a difference of 1 during any stage is obtained, then the entire sequence of natural numbers between the smallest and biggest natural number in the subsequence S can be completed. In accordance with the present disclosure, once centers of possibly broken discs are detected, it is imperative to ascertain that the disc was indeed there. This in turn implies that some part of rediscovered ellipse’s contour overlaps with the actual contour of the overall insulator. Hence, the predicted ellipse’s contour is superimposed onto the output of canny edge detector output in step 204. If
13

there is some edge overlap between the two, whose size (in terms of edge points) is at least 10% of the length of ellipse contour, then it is predicted that a broken disc is indeed present. The broken disc is then highlighted by joining the overlapping edge segments with neighboring edges using Kovesi’s method used in step 204.

[0047] In an embodiment, at step 212, the validated candidate discs of cap insulators from step 210 are scrutinized to identify damaged discs based on inter-center distances between the centers of successive elliptical shapes that may be provided as an output RGB frame at step 214. FIG.6 illustrates detection of a damaged disc 604 in accordance with an embodiment of the present disclosure from elliptical shapes 606 identified pertaining to candidate discs 602 of a cap insulator. In accordance with an embodiment of the present disclosure, a proposed method for predicting location of broken discs is as follows –

List all the centers of ellipses detected post the region growing step 210.
Find the line/axis of alignment of all ellipses.
Find the distance between the centers of two successive ellipses, along the line of alignment.
Find the maximum distance between the set of inter-center distances found so far.
Normalize all the distances with respect to the minimum distance, to get fractional distances.
Round off the fractional distances to the nearest integer, thus getting integral distances, and the corresponding set, S.
Sort the set S.
while size of successive difference set ≠ 1 do
Find the difference between successive numbers in the set, S.
Remove the duplicate differences i.e. form a list of unique natural numbers.
Replace the original set S with the set of unique successive differences.
end While
Using the final minimum distance between any two ellipse centers, complete the sequence between minimum and maximum distances.
If Completed sequence is different and bigger than the original set, s, then
//Newly added numbers to S correspond to probable disc locations.
For each newly added number to S
Multiply the number with the minimum distance / normalization distance. 14

Add the distance to the center coordinates of the lowest ellipse on the axis of alignment, to predict a broken disc.
end for
end if
[0048] The above method of the present disclosure may fail if any damaged disc is the first or last disc in the string. Such discs may be located by extending the step of distance addition, to adding distance on both sides of the current ellipse’s center, along the line of alignment. Further, such predictions must be checked for edge overlap at Canny edge detection stage (step b) of step 204 since at this stage contour of all discs in most scenarios are detected, even as a sequence of smaller edges. Accordingly, if an ellipse in a predicted location does have a corresponding edge overlap in the Canny edge output, it may be shortlisted as one of the broken discs.

[0049] Experimental Analysis of the method of the present disclosure vis-à-vis a method known in the art.

Data Collection
A dataset having about 1014 frames were collected for the study, mostly via flying Unmanned Aerial Vehicles (UAVs), and also by observing transmission towers from a height, via rooftop of high structures, wherever possible. For UAV-based imaging, a 11 MP f/2.8 120_ FOV wide-lens RGB camera, GoPro Hero3, was mounted on a mini-UAV and used for imaging. The frame rate chosen was 10 fps (low value), to avoid too close views of the insulators. For low-altitude rooftop aerial imaging, a standard Canon 12 MP f/2.8 RGB camera was used.
Result and Analysis
The method of detecting candidate discs of cap insulators of the present disclosure was compared with a known method by Oberweger et al. (2014). In terms of absolute performance, the method of the present disclosure was able to pick up true positive insulator regions in 94.1% of the cases. This is an improved performance over 92.7% TPR (true positive rate) quoted in Oberweger et al. (2014). Further, the improvement occurs without any need to elaborately train for insulator objects, as is done via supervised learning in Oberweger et al. (2014). This makes the method of the present disclosure more easily deployable, with better detection performance. 15

Out of 954 frames in which insulator regions were detected, 41 were false detections. From an angle of binary classification, this amounts to precision of 95.7%, and recall of 90%. The precision shown by method of Oberweger et al. (2014) peaks at around 68%, which is significantly lower than that of the present disclosure. A non-trivial part of the dataset used had low-contrast aerial images of insulators. Even in such photometric conditions, the method of the present disclosure was able to detect insulator regions fairly well thereby proving robustness of the method of the present disclosure. In terms of detection of broken discs/caps in specific insulators, the method of the present disclosure was able to locate such discs with 100% accuracy. However, data available for broken insulators is small-sized, as has been the problem for other methods as well such as Oberweger et al. (2014). Hence statistics pertaining to detection of broken discs/caps is not very robust.
[0050] The written description describes the subject matter herein to enable any person skilled in the art to make and use the embodiments of the present disclosure. The scope of the subject matter embodiments defined here may include other modifications that occur to those skilled in the art. Such other modifications are intended to be within the scope if they have similar elements that do not differ from the literal language of the claims or if they include equivalent elements with insubstantial differences from the literal language.

[0051] It is, however to be understood that the scope of the protection is extended to such a program and in addition to a computer-readable means having a message therein; such computer-readable storage means contain program-code means for implementation of one or more steps of the method, when the program runs on a server or mobile device or any suitable programmable device. The hardware device can be any kind of device which can be programmed including e.g. any kind of computer like a server or a personal computer, or the like, or any combination thereof. The device may also include means which could be e.g. hardware means like e.g. an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or a combination of hardware and software means, e.g. an ASIC and an FPGA, or at least one microprocessor and at least one memory with software modules located therein. Thus, the means can include both hardware means and software means. The method embodiments described herein could be implemented in hardware and software. The device may also include software means. Alternatively, the embodiments of the present disclosure may be implemented on different hardware devices, e.g. using a plurality of CPUs.
16

[0052] The embodiments herein can comprise hardware and software elements. The embodiments that are implemented in software include but are not limited to, firmware, resident software, microcode, etc. The functions performed by various modules comprising the system of the present disclosure and described herein may be implemented in other modules or combinations of other modules. For the purposes of this description, a computer-usable or computer readable medium can be any apparatus that can comprise, store, communicate, propagate, or transport the program for use by or in connection with the instruction execution system, apparatus, or device. The various modules described herein may be implemented as software and/or hardware modules and may be stored in any type of non-transitory computer readable medium or other storage device. Some non-limiting examples of non-transitory computer-readable media include CDs, DVDs, BLU-RAY, flash memory, and hard disk drives.

[0053] Further, although process steps, method steps, techniques or the like may be described in a sequential order, such processes, methods and techniques may be configured to work in alternate orders. In other words, any sequence or order of steps that may be described does not necessarily indicate a requirement that the steps be performed in that order. The steps of processes described herein may be performed in any order practical. Further, some steps may be performed simultaneously.

[0054] The preceding description has been presented with reference to various embodiments. Persons having ordinary skill in the art and technology to which this application pertains will appreciate that alterations and changes in the described structures and methods of operation can be practiced without meaningfully departing from the principle, spirit and scope.
17

WE CLAIM:
1. A method comprising:

detecting edge segments representing curves, from at least one image of interest; and
identifying elliptical shapes from pairs of curves that satisfy one or more pre-defined pairing conditions that result in an ellipse, the pairs of curves being identified from the detected edge segments.
2. The method of claim 1, wherein the one or more pre-defined pairing conditions comprise:

each edge segment in the pairs of curves is an approximated set of at least three or more line segments to ensure at least a degree of curvature;
the pair of curves, when extrapolated towards each other converge to result in a convex region; and
concave portions of the pairs of curves face each other.
3. The method of claim 1, wherein identifying elliptical shapes is based on least squares optimization method for ellipse fitting wherein input to the least squares optimization method comprises the pairs of curves and output thereof comprises parameters for identifying elliptical shapes.

4. The method of claim 1 further comprising filtering the elliptical shapes to obtain candidate discs of cap insulators from the at least one image of interest based on pattern selection for linearly oriented shapes.

5. The method of claim 4, wherein filtering the elliptical shapes is based on:

the elliptical shapes having same size;
orientation of the elliptical shapes; and
centers of the elliptical shapes being aligned along a line perpendicular to the orientation.
6. The method of claim 4 further comprising validating the filtered elliptical shapes as the candidate discs of cap insulators based on region growing method.
18

7. The method of claim 6 further comprising detecting damaged candidate discs of cap insulators based on inter-center distances between the centers of successive elliptical shapes.

8. A system comprising:

one or more data storage devices operatively coupled to the one or more processors and configured to store instructions configured for execution by the one or more processors to:
detect edge segments representing curves, from at least one image of interest; and
identify elliptical shapes from pairs of curves that satisfy one or more pre-defined pairing conditions that result in an ellipse, the pairs of curves being identified from the detected edge segments.
9. The system of claim 8, wherein the one or more pre-defined pairing conditions comprise:

each edge segment in the pairs of curves is an approximated set of at least three or more line segments to ensure at least a degree of curvature;
the pair of curves, when extrapolated towards each other converge to result in a convex region; and
concave portions of the pairs of curves face each other.
10. The system of claim 8, wherein the one or more processors are further configured to identify elliptical shapes based on least squares optimization method for ellipse fitting wherein input to the least squares optimization method comprises the pairs of curves and output thereof comprises parameters for identifying elliptical shapes.

11. The system of claim 8, wherein the one or more processors are further configured to filter the elliptical shapes to obtain candidate discs of cap insulators from the at least one image of interest based on pattern selection for linearly oriented shapes.

12. The system of claim 11, wherein the one or more processors are further configured to filter the elliptical shapes based on:

the elliptical shapes having same size;
orientation of the elliptical shapes; and 19

centers of the elliptical shapes being aligned along a line perpendicular to the orientation.
13. The system of claim 11, wherein the one or more processors are further configured to validate the filtered elliptical shapes as the candidate discs of cap insulators based on region growing method.

14. The system of claim 13, wherein the one or more processors are further configured to detect damaged candidate discs of cap insulators based on inter-center distances between the centers of successive elliptical shapes.