Download In this paper they have proposed a new method for shape

Shape Recognition Using MATLAB
Bhupesh Anejaa, Arushi Bharadwajb, Shivang Tripathib, Sudhanshu Joshib
a- Assistant Professor, Instrumentation & Control Department, JSSATE,
c- Student, Mechanical Engineering Department, JSSATE, Noida
Abstract
Shape recognition is one of the key aspects in Computer Vision. From different point of
views the problems of object recognition have been resolved and some of the
modification in the recognition technique is still going on. This is the main reason that
shape recognition is used in a huge number of applications with important challenges
such as noise, degradations. In this paper a number of shapes recognition techniques
have been defined from which researcher can get an idea for modified efficient
techniques.
SHAPE RECOGNITION USING MATLAB
LITERATURE REVIEW
I. INTRODUCTION
Shape analysis is mainly the automatic analysis of geometric shapes, for example a computer
is used to detect the similarly shaped objects in a database or parts that fit together. For a
computer to automatically analyse and process geometric shapes, the objects have to be
represented in a digital form. Most commonly a boundary representation is used to describe
the object with its boundary. However, representations which are based on volume (e.g.
constructive solid geometry) or representations based on point (point clouds) can be used to
represent shape.
The distinguishing is based on some characteristic of the shapes. MATLAB function region
props can automatically calculate some properties of the input image, such as the major axis
length, minor axis length and extent.
The calculation of value extent is done by the rate of minimum enclosing rectangle’s area of
the image and the image’s area. The minimum enclosing rectangle for square and rectangle
are themselves, therefore the extension values for rectangle and square are both 1. Besides,
the major axis length and minor axis length of square are the same, however, for rectangle
these values can be different. Therefore, square and rectangle can be detected now.
II. LITERATURE REVIEW
The basic idea behind doing a literature survey is to gain knowledge regarding the related
work. As was in our case, several research paper were taken into consideration and studied.
Stress is laid to summarize the concept of different authors who has worked in this field.
In the past decades, several three-dimensional face recognition algorithms have been
designed, and evaluated. They vary widely in theory, tools, and methods. Here we propose a
new 3D face recognition algorithm, entirely developed in MATLAB , whose framework
totally comes from differential geometry. The first step being extraction of 17 soft-tissue
landmarks relying on geometrical properties of facial shape.
Enrico Vezzetti, Federica Marcolin, Giulia Fracastoro [1] : Vezetti and Marcolin have made
use of derivatives, coefficients of the fundamental forms, principal, mean, and Gaussian
curvatures, and shape and curvedness indices. Then, a set of geodesic and Euclidean
distances, together with nose volume and ratios between geodesic and Euclidean distances,
has been computed and summed in a final score, used to compare faces. The highest
contribution of this work is that its theoretical basis is differential geometry with its various
descriptors, which is something totally different in the field.
Xiang Bai, Xingwei Yang[2]: Due to problems encountered like distortion, noise,
segmentation errors, overlap, and occlusion of objects in digital images, it is almost
impossible to extract a complete object contour or to segment the objects as a whole.
However, in some cases parts of contours can be correctly reconstructed either by performing
edge grouping or as parts of boundaries of segmented regions. Hence, recognition of objects
based on their contours seems to be a promising research direction. The main contribution of
the paper published by Xiang Bai and Yang is a system for detecting and recognizing of
contour parts in digital images. Detection and recognition, both, are based on shape
correlation of contour parts. For each contour part produced by contour grouping, they have
used shape similarity to get the most similar contour parts in a database of known contour
segments. A shape-based classification of the retrieved contour parts then performs
simultaneous detection and recognition. Complete contours of known objects are parsed
using discrete curve evolution. Then, their representation is constructed that is invariant to
scaling, rotation, and translation.
Rong-Xiang Hu , Wei Jia [3] Xiang Hu and David Zhang proposed hand shape recognition
method which they named as Coherent Distance Shape Contexts (CDSC). It is based on two
classical shape representations, i.e., Shape Contexts (SC) and Inner-distance Shape Contexts
(IDSC). CDSC is capable of capturing discriminative features from hand shape. It can very
well deal with the inexact correspondence problem of hand landmark points. Specifically, it
can extract features mainly from the contour of fingers. For the purpose of verifying the
effectiveness of CDSC, they create a new image database containing 4000 grayscale left
hand images of 200 subjects, on which CDSC has achieved the accurate identification rate of
99.60% for identification and the Equal Error Rate of 0.9% for verification. These are
comparable with the state-of-the-art hand shape recognition methods.
Shefali Sharma[4]:A multimodal biometric system for personal identity verification was
proposed using hand shape and hand geometry in the paper published by Shefali Sharma and
others. Shape and geometry features were derived with the help of only the contour of the
hand image for which only one image acquisition device is sufficient. Processing was done
with respect to a stable reference point at the wrist line which is more stable as compared to
the centroid against the finger rotation and peaks and valleys determination. Two shape based
features were extracted. This was done by using the distance and orientation of each point of
hand contour with respect to the reference point followed by wavelet decomposition to
reduce the dimension. Seven distances were used to encode the geometrical information of
the hand. Shape and geometry based features were fused at score levels and their
performances evaluated using standard ROC curves between false acceptance rate, true
acceptance rate, equal error rate and decidability index. Many similarity measures were used
to examine the accuracy of the introduced method. Performance of system is analysed for
shape based (distance and orientation) and geometrical features individually as well as for all
possible combinations of feature and score level fusion. The proposed features and fusion
methods are studied over two hand image datasets, (1) JUET contact database of 50 subjects
having 10 templates each and (2) IITD contactless dataset of 240 subjects with 5 templates
each. The proposed method outperformed other approaches with the best 0.31% of EER.
Mats Lind , Ned Bingham[5]:Previous studies on 3D shape perception showed a general
inability to visually perceive metric shape. In line with this, studies of object recognition
showed that only qualitative differences, not quantitative or metric ones can be used
effectively for object recognition. Bingham and Lind (2008) found that large perspective
changes (P45) allow perception of metric shape and Lee and Bingham (2010) found that this,
in turn, allowed accurate feedforward reaches-to-grasp objects varying in metric shape. They
investigated whether this information allowed accurate and effective recognition of objects
that vary with respect to metric shape. Both judgment accuracies (d0) and reaction times
confirmed that, with the availability of visual information in large perspective changes,
recognition of objects using quantitative was equivalent in accuracy and speed of judgments.
The ability to recognize objects based on their metric shape is, therefore, a function of the
availability or unavailability of requisite visual information. These issues and results were
discussed in the context of the Two Visual System hypothesis of Milner and Goodale (1995,
2006)
Anuj Srivastav , Turaga and Kurtek[6]:In this paper they have summarized advances in
shape analysis and shape-based activity recognition problems with a focus on techniques that
use tools from differential geometry and statistics. They started with general goals and
challenges faced in shape analysis, followed by a summary of the basic ideas, strengths and
limitations, and applications of different mathematical representations used in shape analyses
of 2D and 3D objects. These representations included point sets, curves, surfaces, level sets,
deformable templates, medial representations, and other feature-based methods. They
discussed some common choices of Riemannian metrics and computational tools used for
evaluating geodesic paths and geodesic distances for several of these shape representations.
They also discussed the use of Riemannian frameworks in statistical modeling of variability
within shape classes. Next were the models and algorithms for activity analysis from various
perspectives. They discussed how mathematical representations for human shape and its
temporal evolutions in videos lead to analyses over certain special manifolds, various choices
of shape features, and parametric and non-parametric models for shape evolution, and how
these choices lead to appropriate manifold-valued constraints. They also discussed
applications of these methods in gait-based biometrics, action recognition, and video
summarization and indexing. For reader convenience, a short overview of the relevant tools
from geometry and statistics has been provided.
Mohammad Reza Daliri and Vincent Torre [7]: In this paper they have proposed a new
method for shape recognition and retrieval. The suggested algorithm was based on several
steps. The algorithm analyses the contour of pairs of shapes. Their contours are recovered and
represented by a pair of N points obtained by linear interpolation. Given two points ‘pi’ and
‘qj’ from the two shapes the cost of their matching is evaluated by using the shape context
and by using dynamic programming the best matching between the point sets is obtained.
Dynamic programming not only recovers the best matching, but also identifies occlusions,
i.e. points in the two shapes which cannot be properly matched. Given the correspondence
between the two point sets, the two contours are aligned using Procrustes analysis. After
alignment, each contour is transformed into a string of symbols and a modified version of edit
distance is used to compute the similarity between strings of symbols. Finally, recognition
and retrieval are obtained by a simple nearest-neighbour procedure. The algorithm has been
tested on a large set of shape databases (Kimia, MPEG-7, natural silhouette database, gesture
database, marine database, Swedish leaf database, diatom database, ETH-80 3D object
database) providing performances for both in recognition and in retrieval superior to most of
previously proposed approaches.
Mark S Nixon[8]:In this paper they have introduced a new multi scale Fourier-based object
description in 2-D space using a low-pass Gaussian filter (LPGF) and a high-pass Gaussian
filter (HPGF), separately. Using the LPGF at different scales (standard deviation) represents
the inner and central part of an object more than the boundary. On the other hand using the
HPGF at different scales represented the boundary and exterior parts of an object more than
the central part. Their algorithms were also organized to achieve size, translation and rotation
invariance. Evaluation indicates that representing the boundary and exterior parts more than
the central part using the HPGF performs better than the LPGF-based multi scale
representation, and in comparison to Zernike moments and elliptic Fourier descriptors with
respect to increasing noise. Multi scale description using HPGF in 2-D also outperformed
wavelet transform-based multi scale contour Fourier descriptors and performed similar to the
perimeter descriptors without any noise.
III. CONCLUSION
This paper presents a short description of various shape recognition techniques in order to
make familiar with the object recognition in image processing. These techniques are based on
a number of shape descriptors and can be used to evolve out a modified method of shape
recognition.
REFERENCES
[1]. Enrico Vezzetti, Federica Marcolin, Giulia Fracastoro “3D face recognition:A automatic
strategy based on geometrical desciptors and landmarks.
[2]. Xiang Bai, Xingwei Yang, Longin Jan Latecki “Detection and recognition of contour
parts based on shape similarity”
[3]. Rong-Xiang Hu , Wei Jia , David Zhang , Jie Gui , Liang-Tu Song “Hand shape
recognition based on coherent distance shape contexts”.
[4]. Shefali Sharma , Shiv Ram Dubey , Satish Kumar Singh , Rajiv Saxena , Rajat Kumar
Singh“Identity verification using shape and geometry of human hands”.
[5]. Young-Lim Lee , Mats Lind , Ned Bingham , Geoffrey P. Bingham,” Object recognition
using metric shape”.
[6]. Anuj Srivastava , Pavan Turaga , Sebastian Kurtek “On advances in differentialgeometric approaches for 2D and 3D shape analyses and activity recognition”.
[7]. Mohammad Reza Daliri, Vincent Torre. “Robust symbolic representation for shape
recognition and retrieval”
[8]. Cem Direkoglu, Mark S. Nixon “Shape classification via image-based multiscale
description”.