<?xml version="1.0" encoding="UTF-8"?><!DOCTYPE article  PUBLIC "-//NLM//DTD Journal Publishing DTD v3.0 20080202//EN" "http://dtd.nlm.nih.gov/publishing/3.0/journalpublishing3.dtd"><article xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink" dtd-version="3.0" xml:lang="en" article-type="research article"><front><journal-meta><journal-id journal-id-type="publisher-id">AJCM</journal-id><journal-title-group><journal-title>American Journal of Computational Mathematics</journal-title></journal-title-group><issn pub-type="epub">2161-1203</issn><publisher><publisher-name>Scientific Research Publishing</publisher-name></publisher></journal-meta><article-meta><article-id pub-id-type="doi">10.4236/ajcm.2013.31A002</article-id><article-id pub-id-type="publisher-id">AJCM-30731</article-id><article-categories><subj-group subj-group-type="heading"><subject>Articles</subject></subj-group><subj-group subj-group-type="Discipline-v2"><subject>Physics&amp;Mathematics</subject></subj-group></article-categories><title-group><article-title>
 
 
  The &lt;i&gt;m&lt;/i&gt;-Point Quaternary Approximating Subdivision Schemes
 
</article-title></title-group><contrib-group><contrib contrib-type="author" xlink:type="simple"><name name-style="western"><surname>hahid</surname><given-names>S. Siddiqi</given-names></name><xref ref-type="aff" rid="aff1"><sup>1</sup></xref><xref ref-type="corresp" rid="cor1"><sup>*</sup></xref></contrib><contrib contrib-type="author" xlink:type="simple"><name name-style="western"><surname>Muhammad</surname><given-names>Younis</given-names></name><xref ref-type="aff" rid="aff2"><sup>2</sup></xref><xref ref-type="corresp" rid="cor1"><sup>*</sup></xref></contrib></contrib-group><aff id="aff1"><addr-line>Department of Mathematics, University of the Punjab, Lahore, Pakistan</addr-line></aff><aff id="aff2"><addr-line>Centre for Undergraduate Studies, University of the Punjab, Lahore, Pakistan</addr-line></aff><author-notes><corresp id="cor1">* E-mail:<email>shahidsiddiqiprof@yahoo.co.uk(HSS)</email>;<email>younis.pu@gmail.com(MY)</email>;</corresp></author-notes><pub-date pub-type="epub"><day>30</day><month>04</month><year>2013</year></pub-date><volume>03</volume><issue>01</issue><fpage>6</fpage><lpage>10</lpage><history><date date-type="received"><day>January</day>	<month>18,</month>	<year>2013</year></date><date date-type="rev-recd"><day>February</day>	<month>19,</month>	<year>2013</year>	</date><date date-type="accepted"><day>March</day>	<month>11,</month>	<year>2013</year></date></history><permissions><copyright-statement>&#169; Copyright  2014 by authors and Scientific Research Publishing Inc. </copyright-statement><copyright-year>2014</copyright-year><license><license-p>This work is licensed under the Creative Commons Attribution International License (CC BY). http://creativecommons.org/licenses/by/4.0/</license-p></license></permissions><abstract><p>
 
 
   In this article, the objective is to introduce an algorithm to produce the quaternary m-point (for any integer m&gt;1) approximating subdivision schemes, which have smaller support and higher smoothness, comparing to binary and ternary schemes. The proposed algorithm has been derived from uniform B-spline basis function using the Cox-de Boor recursion formula. In order to determine the convergence and smoothness of the proposed schemes, the Laurent polynomial method has been used. 
 
</p></abstract><kwd-group><kwd>Cox-De Boor Recursion Formula; Quaternary; Approximating Subdivision Schemes; Convergence and Smoothness</kwd></kwd-group></article-meta></front><body><sec id="s1"><title>1. Introduction</title><p>Until a few years ago all the work in the area of univariate subdivision was limited to consider just binary (Chaikin [<xref ref-type="bibr" rid="scirp.30731-ref1">1</xref>]; Dyn et al. [2,3]; Siddiqi and Younis [<xref ref-type="bibr" rid="scirp.30731-ref4">4</xref>]; Beccari et al. [<xref ref-type="bibr" rid="scirp.30731-ref5">5</xref>]) and ternary (Hassan and Dodgson [<xref ref-type="bibr" rid="scirp.30731-ref6">6</xref>]; Hassan et al. [<xref ref-type="bibr" rid="scirp.30731-ref7">7</xref>]; Ko et al. [<xref ref-type="bibr" rid="scirp.30731-ref8">8</xref>]; Mustafa et al. [<xref ref-type="bibr" rid="scirp.30731-ref9">9</xref>]) scenarios. In recent time, some proposals of quaternary subdivision schemes have introduced new interest in the era of subdivision, showing the possibility of treating refinement schemes with arity other than two or three.</p><p>Since subdivision schemes propose efficient iterative algorithms to produce the smooth curves and surfaces from a discrete set of control points by subdividing them according to some refining rules, recursively. These refining rules are very helpful and useful for the creation of smooth curves and surfaces in computational geometry and geometric designing due to their wide range of applications in many areas like engineering, medical science and image processing etc.</p><p>In this article an algorithm has been introduced to produce the quaternary <img src="2-1100220\36f22e38-6d68-4144-a993-14ba4ae304bb.jpg" />point (for any integer<img src="2-1100220\ca79030b-9d63-45f3-8a03-abd3bf195237.jpg" />) approximating subdivision schemes. This algorithm has been developed using the Cox-de Boor recursion formula, in the form of uniform B-spline blending functions to produce piecewise polynomials of order <img src="2-1100220\6f8f6f5e-c84b-4082-86f3-268dad86478e.jpg" /> over the interval <img src="2-1100220\883c096c-024a-49c0-895d-0567501333e6.jpg" /> (for detail, see Section 2).</p><p>The quaternary subdivision scheme can be defined in terms of a mask consisting of a finite set of non-zero coefficients <img src="2-1100220\588f4183-77d1-43bb-a1d2-e4b864e27c7c.jpg" /> as follows</p><p><img src="2-1100220\23050019-0e4c-45c0-b96a-a09d6c5527db.jpg" /></p><p>The formal definitions and the notion for the convergence analysis of the quaternary subdivision scheme are as follows:</p><p>The quaternary convergent subdivision scheme <img src="2-1100220\d135bf37-e221-4a02-b71b-9fe438d2ea8f.jpg" /> with the corresponding mask <img src="2-1100220\1c029ff0-c28e-4007-8bdc-05921cd97775.jpg" /> necessarily satisfies</p><p><img src="2-1100220\910ed386-4548-4a02-8aa1-e0f6e2baaefe.jpg" /></p><p>It follows that the symbol of a convergent subdivision scheme satisfies the conditions <img src="2-1100220\b1b8de7f-ffd4-4f31-9431-d6b4bc1c02aa.jpg" /> and<img src="2-1100220\c1308032-1491-4d94-8dcb-40cc91bc5f45.jpg" />for<img src="2-1100220\cc180452-2560-4bdd-ab1a-e09cbf17a532.jpg" />.</p><p>Introducing a symbol called the Laurent polynomial</p><p><img src="2-1100220\74276503-5ccc-48d8-8e64-865534fa6bc7.jpg" /></p><p>of a mask <img src="2-1100220\3fccc85f-7405-4e8e-8e03-65dca71dcd63.jpg" /> with finite support. In view of Dyn [<xref ref-type="bibr" rid="scirp.30731-ref3">3</xref>], the sufficient and necessary conditions for a uniform convergent scheme are defined as follows.</p><p>A subdivision scheme <img src="2-1100220\84736cdf-ba77-42a9-bc69-2ed93206e591.jpg" /> is uniform convergent if and only if there is an integer<img src="2-1100220\6e617d09-8231-4872-9648-5e29e385296b.jpg" />, such that</p><p><img src="2-1100220\b96a9076-86a6-4e3e-969f-aa9f2bb8130e.jpg" /></p><p>subdivision <img src="2-1100220\3a4884b1-17f0-4abe-aedc-5ac949cb5d91.jpg" /> with symbol <img src="2-1100220\bbb7d393-b8ea-47e8-8664-fbe576aeb398.jpg" /> is related to S with symbol<img src="2-1100220\81a8966e-7385-4f6c-b850-62ccf7d386ff.jpg" />, where <img src="2-1100220\1ceeeeea-c2fc-47f9-adde-f9b2273fca0b.jpg" /> and satisfying the property</p><p><img src="2-1100220\8c7c35c8-f293-47e2-97f6-3c70cec86a65.jpg" /></p><p>where <img src="2-1100220\65a6d914-6841-4e91-85ce-4f9c0c5eb6dc.jpg" /> and</p><p><img src="2-1100220\086a2227-3bff-4b87-9632-bc109dd0946f.jpg" />The norm <img src="2-1100220\fff0d4f7-bd69-43a0-a728-5afba41d6ed0.jpg" /></p><p>of a subdivision scheme <img src="2-1100220\e4138b4d-0669-4bad-9b8a-260b8e888344.jpg" /> with a mask <img src="2-1100220\6398bd85-6c86-42d6-88af-a4d2b43ac9d3.jpg" /> is defined by</p><p><img src="2-1100220\e7408f8e-b3f6-4448-a60c-2ae07cc99d05.jpg" /></p><p>and</p><p><img src="2-1100220\cc677a5c-31b8-4b0e-ba00-8470646e5316.jpg" /></p><p>where</p><p><img src="2-1100220\283b86e0-8b11-4600-b8f8-599db5b1cf53.jpg" /></p><p>where</p><p><img src="2-1100220\4084c198-4dc6-42cf-a776-cdfbcdd64766.jpg" /></p><p>and</p><p><img src="2-1100220\942d8778-2eae-439d-95b1-9e501cc2c1a9.jpg" /></p><p>The paper is organized as follows, in Section 2 the algorithm to construct <img src="2-1100220\02b8dfd9-7f8b-44d0-b897-55a375390940.jpg" />-point (for any integer<img src="2-1100220\abca0952-9648-4a4c-8990-3d52845a1445.jpg" />) quaternary schemes has been introduced. Three examples are considered to produce the masks of 2-point (corner cutting), 3-point and 4-point schemes in the same section. In Section 3, the polynomial reproduction property has been discussed. The conclusion is drawn in Section 4.</p></sec><sec id="s2"><title>2. Construction of the Algorithm</title><p>In this section, an algorithm has been constructed to produce the quaternary <img src="2-1100220\bffd838c-d21e-4e0c-8663-5a8e30529967.jpg" />-point approximating subdivision schemes using the uniform B-spline basis functions and the Cox-de Boor recursion relation. The Cox-de Boor recursion relation, in view of Buss [<xref ref-type="bibr" rid="scirp.30731-ref10">10</xref>], can be defined as follows:</p><p>The recursion relation is the generalization to B-spline of degree <img src="2-1100220\522c4cfa-8654-4e30-b23c-ab644485c4e9.jpg" /> (or of order<img src="2-1100220\87ccc5bf-1acf-4676-9273-ea6e950abefd.jpg" />, i.e.,<img src="2-1100220\21ebfac8-c5c2-4b33-92ab-acaf05dae101.jpg" />). For this, consider <img src="2-1100220\41e5caf8-a624-4e85-86c5-c0bb4c650e20.jpg" /> to be a set of <img src="2-1100220\0820e585-51b6-4e79-96fd-bcaf88d5bde4.jpg" /> non-decreasing real numbers in such a way that<img src="2-1100220\6a52e800-1c38-4d77-8c93-d8ecc8377a37.jpg" />. The values<img src="2-1100220\90bec16c-0ed7-4e90-9694-4c9e3d87d46b.jpg" />’s, not necessarily uniformly spaced, are called knots of non-uniform spline and the set <img src="2-1100220\8a4a3a5b-debc-4a46-85a2-5798101f1d6d.jpg" /> is called knot vector. The uniform B-splines are just the special case of non-uniform B-splines in which the knots are equally spaced such that <img src="2-1100220\3d7c52f5-bb98-4ca4-8e8e-b075373f1fc1.jpg" /> is a constant for <img src="2-1100220\1885c4dc-fad7-497d-81a9-8c71887ef48f.jpg" /> (i.e.,<img src="2-1100220\7c3d7562-5d9c-4222-a5ad-07e1371d236a.jpg" />) . Note that the blending functions <img src="2-1100220\59791ba8-74e0-48c9-9159-e28b634c7d98.jpg" /> of order <img src="2-1100220\d4a5b8d2-e392-4337-bbec-ea786c228380.jpg" /> depend only on the knot positions and are defined by induction on <img src="2-1100220\1d7e44e6-58c6-4ff7-89c6-f9825aa6e0a5.jpg" /> as follows.</p><p>First, for <img src="2-1100220\67d16a47-9f86-4faf-852b-f1139ffa7636.jpg" /> let</p><p><img src="2-1100220\5531eabd-a56b-4715-8c3d-b7cdadd5150c.jpg" /></p><p>Second for<img src="2-1100220\e98d7cd6-caa6-4277-bf83-1627cbbfaa9a.jpg" />. Setting<img src="2-1100220\8a2b9cc9-0b56-43b8-be26-ff1cadbe1797.jpg" />, <img src="2-1100220\ea4d8213-64f6-4df3-802d-01aea3e8d1db.jpg" />is defined by the Cox-de Boor formula as,</p><disp-formula id="scirp.30731-formula49030"><label>(1)</label><graphic position="anchor" xlink:href="2-1100220\fa704cd6-880f-435d-a4ae-20634a0b93f1.jpg"  xlink:type="simple"/></disp-formula><p>The form of above recursive formulas for the blending function immediately implies that the functions <img src="2-1100220\d34e1849-38ae-49e2-949e-a1a23bc4eac7.jpg" /> are piecewise polynomials of degree <img src="2-1100220\adb3995a-9745-4f59-a2a4-e41cd66c1bb9.jpg" /> and that the breaks between pieces occur at the knots<img src="2-1100220\ff0695cd-ef6b-4120-95d6-d93596fc89f2.jpg" />.</p><p>In view of above recursion formula, the Uniform Bspline blending functions <img src="2-1100220\faffe130-e879-43c2-b915-b0ad991a8ebe.jpg" /> of order <img src="2-1100220\9519ec5c-3051-4968-bc9b-30dbd6c78332.jpg" /> over the interval<img src="2-1100220\9c9ace17-6d84-4342-b6bb-9957d3d422d1.jpg" />, together with the properties [<xref ref-type="bibr" rid="scirp.30731-ref10">10</xref>], can be defined in Equation (2).</p><p>The blending functions must satisfy the following properties:</p><p>• The blending functions are translates of each other, that is,<img src="2-1100220\c5935acc-b328-4ddf-b360-73570bd1c76f.jpg" />.</p><p>• The blending functions are a partition of unity, that is,<img src="2-1100220\30d778a4-34ef-4040-9658-afac9df885f3.jpg" />.</p><p>• <img src="2-1100220\8f53b4ae-d547-4eee-ba95-479f8a269ada.jpg" />for all t.</p><p>• The functions <img src="2-1100220\ba3e0e97-c4a3-49f3-899e-c0ca771bbf7c.jpg" /> have continuous <img src="2-1100220\ee3f5047-c32a-40e4-a1a9-e5356c7a97e0.jpg" /> derivatives, that is, they are <img src="2-1100220\b201a06d-dc18-4018-af01-6c7aa153a5e6.jpg" />-continuous.</p><disp-formula id="scirp.30731-formula49031"><label>(2)</label><graphic position="anchor" xlink:href="2-1100220\4294bbaf-0888-4899-94b8-53ba47fc82d0.jpg"  xlink:type="simple"/></disp-formula><p>The masks <img src="2-1100220\94e1e46c-5979-4c06-84ec-f83060a9937b.jpg" /> of the proposed quaternary <img src="2-1100220\978f3290-c76a-4b63-afbb-273fc872365b.jpg" />-point scheme can be calculated using the following recurrence relation</p><disp-formula id="scirp.30731-formula49032"><label>(3)</label><graphic position="anchor" xlink:href="2-1100220\b215c8df-3133-44fb-b10d-3f4209a9c197.jpg"  xlink:type="simple"/></disp-formula><p>where <img src="2-1100220\ab124e4c-f874-4915-9860-1c069f27359e.jpg" /> is a uniform B-spline basis function of degree<img src="2-1100220\77657f84-461c-47cd-95a3-65c89169d080.jpg" />. In the following, some examples are considered to produce the masks of 2-point, 3-point and 4-point quaternary approximating schemes after setting <img src="2-1100220\15b43e50-6405-414c-b43d-3852b225e181.jpg" /> and 4, respectively, in recurrence relation (3).</p><p>The 2-point scheme: To obtain the mask of quarternary 2-point scheme, set <img src="2-1100220\4e3ea695-1922-4aaa-9a03-85180f731fab.jpg" /> in above relation (3). It may be noted that the linear uniform B-spline basis function <img src="2-1100220\01b44bdc-8b14-42a8-ae83-3a296efc3d95.jpg" /> produces the mask of 2-point quarternary scheme (which is also called corner cutting scheme). Thus 2-point scheme (after adjusting the mask) to refine the control polygon is defined as follows:</p><disp-formula id="scirp.30731-formula49033"><label>(4)</label><graphic position="anchor" xlink:href="2-1100220\6aea7e34-be28-4723-acae-4139ab912ff9.jpg"  xlink:type="simple"/></disp-formula><p>Now, the convergence and smoothness of the proposed 2-point scheme can be analyzed using the Laurent polynomial method introduced by Tang et al. [<xref ref-type="bibr" rid="scirp.30731-ref11">11</xref>].</p><p>Theorem 2.1: The quaternary 2-point approximating subdivision scheme converges and has smoothness<img src="2-1100220\399709db-c73e-46f5-afff-599a82a1a6fd.jpg" />.</p><p>Proof. To prove that the subdivision scheme <img src="2-1100220\8dad2396-8172-4066-a665-7f374aab937b.jpg" /> corresponding to the symbol <img src="2-1100220\cd6b5a6c-45ac-46f2-a2bf-cb2af4ec66d4.jpg" /> is<img src="2-1100220\bbc0d4bd-5c7d-4659-ae0c-f8cacab9d6e5.jpg" />. So, the Laurent polynomial <img src="2-1100220\ffa2691e-234e-420c-a347-dfa28af57c27.jpg" /> for the mask of the scheme can be written as</p><disp-formula id="scirp.30731-formula49034"><label>(5)</label><graphic position="anchor" xlink:href="2-1100220\38b2d6af-d27f-41a2-b7a1-5989e305a981.jpg"  xlink:type="simple"/></disp-formula><p>The Laurent polynomial method is used to prove the smoothness of the scheme to be<img src="2-1100220\a422c516-77f9-4726-aa41-97a855c5b517.jpg" />. Taking</p><p><img src="2-1100220\2818bb3e-39d7-471e-b044-bc6225162e1c.jpg" /></p><p>where</p><p><img src="2-1100220\bd49165a-9e35-4f4f-9f18-42352f36a53e.jpg" /></p><p>and</p><p><img src="2-1100220\d578abeb-4b00-45fd-a263-fba6122e8146.jpg" /></p><p>With a choice of <img src="2-1100220\d514408b-4fca-4e21-93c0-af9fbed8485c.jpg" /> and<img src="2-1100220\5d8af0fe-03c5-44c4-a107-8055925f0eb6.jpg" />, it can be written as</p><p><img src="2-1100220\7f7cbea6-93f4-4c65-bdf0-fa6e96f367d4.jpg" /></p><p>Since the norm of subdivision <img src="2-1100220\94fb72b6-e4a0-47d3-90bf-2931facb6b6b.jpg" /> is</p><p><img src="2-1100220\6b32507f-3ec6-472f-94ec-1cf083b0c4f9.jpg" /></p><p>therefore <img src="2-1100220\af11776b-149b-4762-b308-3a4cafdf5701.jpg" /> is contractive, by Theorem 3, and so <img src="2-1100220\e55f50de-09e6-4dda-9c42-a9ed5c50b59e.jpg" /></p><p>is convergent.</p><p>In order to prove the scheme developed to be<img src="2-1100220\d78c3f9a-55bf-4289-89d3-ee26ce5e5de4.jpg" />, consider <img src="2-1100220\7aced9c4-9307-4584-a8d4-7a25ea08d7a6.jpg" /> and<img src="2-1100220\3a2fee81-940e-4db7-86ce-ab403f8bf2bf.jpg" />; it can be written as</p><p><img src="2-1100220\a611fed3-7579-4973-a7cf-9d98e7db7e77.jpg" /></p><p>Since the norm of subdivision <img src="2-1100220\dc87bcfb-2438-42b7-b8ac-de65f46aa700.jpg" /> is</p><p><img src="2-1100220\18d489f5-9151-48c1-9d2e-3baea6b990d0.jpg" /></p><p>therefore <img src="2-1100220\a688719a-e02a-4f84-b6c4-00391815f1f3.jpg" /> is contractive. Consequently, <img src="2-1100220\35c1c926-c9c4-4c8f-bc46-cc1fa83c3419.jpg" />is convergent and<img src="2-1100220\eb39fca0-e546-4b58-bbb1-37961c2afa94.jpg" />.</p><p>The 3-point scheme: To obtain the mask of quarternary univariate 3-point scheme, set <img src="2-1100220\3eac0a20-31c4-421e-8f36-de291e060ea1.jpg" /> in recursion relation (3). The quadratic uniform B-spline basis functions <img src="2-1100220\3a0ad371-e313-491a-b15b-514711421caf.jpg" /> are obtained. The mask <img src="2-1100220\4a3038be-0371-4f08-b172-353c8ef05a39.jpg" /> of the proposed quaternary 3-point scheme can be calculated from these basis functions. The 3-point scheme (after adjusting the mask), to refine the control polygon, is defined as:</p><disp-formula id="scirp.30731-formula49035"><label>(6)</label><graphic position="anchor" xlink:href="2-1100220\e801b932-b2f6-43d5-a92c-cef274068a52.jpg"  xlink:type="simple"/></disp-formula><p>Theorem 2.2: The quaternary 3-point approximating subdivision scheme converges and has smoothness<img src="2-1100220\f1cb1aa7-bf4c-48ad-8052-964dbf5fe109.jpg" />.</p><p>Proof. The smoothness of the above subdivision scheme can be calculated following the same procedure.</p><p>The 4-point scheme: Now, a 4-point quaternary scheme is presented and masks of the scheme can be calculated from the cubic basis function. After setting <img src="2-1100220\f059195d-f9c4-454b-ac82-b6bf5c5fc1c3.jpg" /> in relation (3) the cubic B-spline basis functions <img src="2-1100220\a7c089ff-315c-4895-a961-9b3dbd575484.jpg" /> can be calculated. Thus, 4-point scheme is defined as follows</p><disp-formula id="scirp.30731-formula49036"><label>(7)</label><graphic position="anchor" xlink:href="2-1100220\6c2b2e59-888f-4467-a056-45d86056d580.jpg"  xlink:type="simple"/></disp-formula><p>Theorem 2.3: The quaternary 4-point approximating subdivision scheme converges and has smoothness<img src="2-1100220\201fed46-da7a-4d64-98af-64f1dd155ae4.jpg" />.</p><p>Proof. The smoothness of the above subdivision scheme can be calculated following the same procedure.</p><p>In the following section the polynomial reproduction property has been discussed.</p></sec><sec id="s3"><title>3. Properties</title><p>The polynomial reproduction property has its own importance. As, the reproduction property of the polynomials up to certain degree <img src="2-1100220\f60205c6-c124-4035-a72f-e8f02f03b657.jpg" /> implies that the scheme has <img src="2-1100220\35dbfd63-5bb8-4101-93e8-1ef8591b1a04.jpg" /> approximation order. For this, polynomial reproduction can be made from initial date which has been sampled from some polynomial function. In view of [<xref ref-type="bibr" rid="scirp.30731-ref12">12</xref>], the polynomial reproduction property of the proposed scheme can be obtained after having the parametrization <img src="2-1100220\fc8d1770-5368-45f2-98e4-114cb777c5a7.jpg" /> and definitions in the following manner.</p><p>Definition 3.1: For quaternary subdivision scheme the parametrization <img src="2-1100220\6cfdf418-26a0-4019-aded-e365325e6867.jpg" /> the corresponding parametric shift and attach the data <img src="2-1100220\d8b8a018-90c4-4bd6-9693-3684207296f0.jpg" /> for <img src="2-1100220\2e25fd6c-f6c1-4b5b-90dc-b074249f0e16.jpg" /> to the parameter values</p><disp-formula id="scirp.30731-formula49037"><label>(8)</label><graphic position="anchor" xlink:href="2-1100220\b4525ddb-c114-4d6a-8fcc-e07241317f77.jpg"  xlink:type="simple"/></disp-formula><p>Definition 3.2: A quaternary subdivision scheme reproduces polynomial of degree <img src="2-1100220\d7323996-5618-48fc-8561-998c708c08fd.jpg" /> if it is convergent and its continuous limit function (for any polynomial<img src="2-1100220\2bd64189-0465-4218-938a-eb5acd76ad47.jpg" />) is equal to <img src="2-1100220\62631500-32dc-46cb-b1ff-c22c8a626f3c.jpg" /> and initial data</p><p><img src="2-1100220\69687772-1285-4019-b9db-7f0a60126de9.jpg" /></p><p>Theorem 3.3: A convergent quaternary subdivision scheme reproduces polynomials of degree <img src="2-1100220\fa687ccd-f4d4-49a2-9e84-104a037bd9fd.jpg" /> with respect to the parametrization defined in (8) if and only if</p><p><img src="2-1100220\143b831e-4ee3-45b8-87a6-4a5a4b92d505.jpg" /></p><p>Proof. The induction over <img src="2-1100220\2342fe99-97d9-4e2c-afbd-28e6042801d3.jpg" /> can be performed to prove this theorem following [<xref ref-type="bibr" rid="scirp.30731-ref12">12</xref>].</p><p>In view of [<xref ref-type="bibr" rid="scirp.30731-ref12">12</xref>], the following proposition helps to find the necessary conditions defined in (9).</p><p>Proposition 3.4: Let <img src="2-1100220\8fbad6d9-619a-457f-8cc0-36ac3d206be1.jpg" /> and<img src="2-1100220\8ebbaa72-63b4-4f99-8ffb-2aa17b49e603.jpg" />. Then a subdivision symbol <img src="2-1100220\466f5d38-53d0-44af-9ee0-778d61b85840.jpg" /> satisfies</p><disp-formula id="scirp.30731-formula49038"><label>(9)</label><graphic position="anchor" xlink:href="2-1100220\dee74ea3-7ab3-4e1f-8f89-c011ad9bda22.jpg"  xlink:type="simple"/></disp-formula><p>if and only if <img src="2-1100220\c0f5a5e5-d1c9-441b-a10a-a83aab95d124.jpg" /> satisfies</p><disp-formula id="scirp.30731-formula49039"><label>(3)</label><graphic position="anchor" xlink:href="2-1100220\b4c11ca9-5b65-40cc-b63b-c4029d598be5.jpg"  xlink:type="simple"/></disp-formula><p>(<img src="2-1100220\0c140f92-5f68-4014-9d53-8fa25220af61.jpg" />derivative of the symbol) which in turn is equivalent to require that <img src="2-1100220\4fa82469-fe70-4258-a691-897c7cf23fd9.jpg" /> for some c(z).</p><p>Proposition 3.5: Let a quaternary subdivision scheme that reproduces polynomial up to degree<img src="2-1100220\f84a8bb8-9f72-4b2f-aee5-7b2e2c434fa2.jpg" />. Then the smoothed scheme <img src="2-1100220\5df915ab-5674-4002-bab7-90ad282f1acd.jpg" /> with the symbol</p><p><img src="2-1100220\afcf7b7f-e96a-44f3-b067-8f7457cc6642.jpg" />satisfies the conditions</p><p><img src="2-1100220\f97a3c57-0ea5-4dc4-9523-5f6097d59131.jpg" /></p><p>and hence generates polynomial of degree<img src="2-1100220\f8f90049-da06-483d-b6f9-1be10500e5fc.jpg" />, but it has only linear reproduction.</p><p>Proof. Following [<xref ref-type="bibr" rid="scirp.30731-ref12">12</xref>], for some Laurent polynomial</p><p><img src="2-1100220\53121394-b251-413a-8711-53be63ee6ed4.jpg" />with<img src="2-1100220\6c01e7cc-29ad-4983-b0af-10ce0ed2c184.jpg" />, we have</p><p><img src="2-1100220\fbb76bbd-cb45-47bc-98fe-399235561160.jpg" /></p><p>and the fact<img src="2-1100220\00486d1e-f05a-4e28-acbd-fc0fadb9638e.jpg" />. Thus, the <img src="2-1100220\71369d54-5ae4-4678-8e00-9971470856b3.jpg" /> derivative of <img src="2-1100220\0971b2a8-8d5d-42e9-9e16-a8c528ea4923.jpg" /> is</p><p><img src="2-1100220\d2f423f4-be2e-4390-b535-0f8aca0b57b3.jpg" /></p><p>and correct parametric shift for <img src="2-1100220\44e083b0-2e5b-46ac-8bfc-8caf77fa46b2.jpg" /> is</p><p><img src="2-1100220\37ca0d94-8c9e-4890-b30b-67e447f552a7.jpg" /></p><p>The <img src="2-1100220\1640f495-0d86-4554-88c5-a54d210f7f73.jpg" /> derivative of <img src="2-1100220\e6ab193f-515f-4e93-8956-131c94d9f26f.jpg" /> is</p><p><img src="2-1100220\0ced5368-8b56-4223-a083-8e1a70b37586.jpg" /></p><p>which produces</p><p><img src="2-1100220\39205402-2276-4222-beac-cdd3a32064df.jpg" /></p><p>after simplification, it can be yielded that</p><p><img src="2-1100220\9a4c0e0c-da76-45c0-aea0-ee48a2a6a47e.jpg" />.</p><p>Hence, it does not reproduce polynomials of degree<img src="2-1100220\631015d8-066d-4f0e-ad37-59e7eac22786.jpg" />.</p></sec><sec id="s4"><title>4. Conclusion</title><p>A quaternary univariate <img src="2-1100220\800e634d-8328-455c-a859-0bb7b2275c58.jpg" />point (for any integer<img src="2-1100220\2503bdeb-e581-4b28-80ae-8e3bc5df15df.jpg" />) approximating subdivision scheme has been developed which generates the smooth limiting curves. The construction of the quaternary scheme is associated with an algorithm of uniform B-spline basis functions developed from the Cox-de Boor recursion formula. The objective is to introduce the quaternary subdivision schemes, which have smaller support and higher smoothness, comparing to binary and ternary schemes. Moreover the polynomial reproduction property has also been discussed.</p></sec><sec id="s5"><title>REFERENCES</title></sec></body><back><ref-list><title>References</title><ref id="scirp.30731-ref1"><label>1</label><mixed-citation publication-type="other" xlink:type="simple">G. M. Chaikin, “An Algorithm for High Speed Curve Generation,” Graph cuts in Computer Vision, Vol. 3, No. 4, 1974, pp. 346-349.</mixed-citation></ref><ref id="scirp.30731-ref2"><label>2</label><mixed-citation publication-type="other" xlink:type="simple">N. Dyn, J. A. Gregory and D. Levin, “A 4-Points Interpolatory Subdivision Scheme for Curve Design,” Computer Aided Geometric Design, Vol. 4, No. 4, 1987, pp. 257-268. doi:10.1016/0167-8396(87)90001-X</mixed-citation></ref><ref id="scirp.30731-ref3"><label>3</label><mixed-citation publication-type="other" xlink:type="simple">N. Dyn, “Tutorials on Multresolution in Geometric Modelling,” In: A. Iske, E. Quak and M. S. Floater, Eds., Summer School Lectures Notes Series: Mathematics and Visualization, Springer, 1995, ISBN: 3-540-43639-1.</mixed-citation></ref><ref id="scirp.30731-ref4"><label>4</label><mixed-citation publication-type="other" xlink:type="simple">S. S. Siddiqi and M. Younis, “Construction of m-Point Approximating Subdivision Schemes,” Applied Mathematics Letters, Vol. 26, No. 3, 2013, pp. 337-343. 
doi:10.1016/j.aml.2012.09.016</mixed-citation></ref><ref id="scirp.30731-ref5"><label>5</label><mixed-citation publication-type="other" xlink:type="simple">C. Beccari, G. Casciola and L. Romani, “A Non-Stationary Uniform Tension Controlled Interpolating 4-Point scheme Reproducing Conics,” Computer Aided Geometric Design, Vol. 24, No. 1, 2007, pp. 1-9. 
doi:10.1016/j.cagd.2006.10.003</mixed-citation></ref><ref id="scirp.30731-ref6"><label>6</label><mixed-citation publication-type="other" xlink:type="simple">M. F. Hassan and N. A. Dodgson, “Ternary and Three Point Univariate Subdivision Schemes,” In: A. Cohen, J.-L. Merrien and L. L. Schumaker, Eds., Curve and Surface Fitting: Sant-Malo, Nashboro Press, Brentwood, 2003, pp. 199-208.</mixed-citation></ref><ref id="scirp.30731-ref7"><label>7</label><mixed-citation publication-type="other" xlink:type="simple">M. F. Hassan, I. P. Ivrissimtzis, N. A. Dodgson and M. A. Sabin, “An Interpolating 4-Point Ternary Stationary Subdivision Scheme,” Computer Aided Geometric Design, Vol. 19, No. 1, 2002, pp. 1-18. 
doi:10.1016/S0167-8396(01)00084-X</mixed-citation></ref><ref id="scirp.30731-ref8"><label>8</label><mixed-citation publication-type="other" xlink:type="simple">K. P. Ko, B.-G. Lee and G. J. Yoon, “A Ternary 4-Point Approximating Subdivision Scheme,” Applied Mathematics and Computation, Vol. 190, No. 2, 2007, pp. 1563-1573. doi:10.1016/j.amc.2007.02.032</mixed-citation></ref><ref id="scirp.30731-ref9"><label>9</label><mixed-citation publication-type="other" xlink:type="simple">G. Mustafa, A. Ghaffar and F. Khan, “The Odd-Point Ternary Approximating Schemes,” American Journal of Computational Mathematics, Vol. 1, No. 2, 2011, pp. 111-118.</mixed-citation></ref><ref id="scirp.30731-ref10"><label>10</label><mixed-citation publication-type="other" xlink:type="simple">S. R. Buss, “3-D Computer Graphics A Mathematical Introduction with OpenGL,” 1st Edition, Cambridge University Press, New York, 2003. 
doi:10.1017/CBO9780511804991</mixed-citation></ref><ref id="scirp.30731-ref11"><label>11</label><mixed-citation publication-type="other" xlink:type="simple">Y. Tang, K. P. Ko and B. G. Lee, “A New Proof of the Smoothness of 4-Point Deslauriers-Dubuc Scheme,” Journal of Applied Mathematics and Computing Vol. 18, No. 1-2, 2005, pp. 553-562.</mixed-citation></ref><ref id="scirp.30731-ref12"><label>12</label><mixed-citation publication-type="other" xlink:type="simple">C. Conti and K. Hormann, “Polynomial Reproduction for Univariate Subdivision Schemes of any Arity,” Journal of Approximation Theory, Vol. 163, No. 4, 2011, pp. 413-437. doi:10.1016/j.jat.2010.11.002</mixed-citation></ref></ref-list></back></article>