From fc4dc3d3a05d6e3c892f80ba7edb83bea626e4b9 Mon Sep 17 00:00:00 2001
From: Jeena <spam@jeenaparadies.net>
Date: Tue, 15 Oct 2013 18:50:49 +0200
Subject: [PATCH] added report

---
 find_lines.py |  28 ++++---
 report.html   | Bin 0 -> 47986 bytes
 report.md     | 210 ++++++++++++++++++++++++++++++++++++++++++++++++++
 3 files changed, 223 insertions(+), 15 deletions(-)
 create mode 100644 report.html
 create mode 100644 report.md

diff --git a/find_lines.py b/find_lines.py
index 13c3a43..3168d70 100755
--- a/find_lines.py
+++ b/find_lines.py
@@ -13,11 +13,8 @@ def extractFeatures(label):
 
 		img = cv2.imread(directory + fn, 0)
 
-		#temp = cv.CreateImage((100,100), cv.CV_8U, 1)
-		#cv.Smooth(img, temp)
-
+		# find edges
 		canny = cv2.Canny(img, 50, 100)
-		color_dst = cv2.cvtColor(img, cv2.COLOR_GRAY2BGR)
 
 		# find colored
 		black_pixels = numpy.count_nonzero(img)
@@ -30,8 +27,11 @@ def extractFeatures(label):
 		try:
 			for line in lines[0]:
 				x1, y1, x2, y2 = line
-				#cv2.line(color_dst, (x1, y1), (x2, y2), cv.RGB(255,0,0), 1, 8)
-				length = int(math.sqrt(math.pow((x1-x2), 2) + math.pow((y1-y2), 2)))
+
+				# Pythagoras
+				a2 = math.pow((x1-x2), 2)
+				b2 = math.pow((y1-y2), 2)
+				length = int(math.sqrt(a2 + b2))
 				lengths.append(length)
 
 				angle = int(math.degrees(math.atan((y1-y2) / (x1-x2))))
@@ -44,13 +44,10 @@ def extractFeatures(label):
 		mid_length = sum(lengths) / lines_count
 		mid_angle = sum(angles) / lines_count
 
-		features.append([[lines_count, mid_length, mid_angle, black_pixels], label])
-
-		#cv2.namedWindow("Original")
-		#cv2.imshow("Original", img)
-
-		#cv2.namedWindow('Lines image ' + fn)
-		#cv2.imshow('Lines image ' + fn, color_dst)
+		features.append([
+			[lines_count, mid_length, mid_angle, black_pixels],
+			label
+		])
 
 	return features
 
@@ -72,7 +69,8 @@ if __name__ == "__main__":
 	tree = KDTree(features)
 	
 	for t in xrange(0, test_count * 2):
-		d, i = tree.query(test_features[t], k=2)
+		d, i = tree.query(test_features[t], k=3)
+		print "-"
 		for j in xrange(0, len(i)):
-			print test_labels[t] + " is predicted to be a " + labels[i[j]] + " j: " + str(i[j]) + " d: " + str(d[j])
+			print test_labels[t] + " is predicted to be a " + labels[i[j]]
 
diff --git a/report.html b/report.html
new file mode 100644
index 0000000000000000000000000000000000000000..c44e3db6ac249905dd1516e2e246cf3c53275ec3
GIT binary patch
literal 47986
zcmeI5`BNRqm8kE}{a4`j^F;S(B=${idqyk*1OfyIkUWkaHUVOl*n}MKUw`j?Umj9f
zRrj*A)NVgVIBwjfsxnWWJbRuzRrmk?pPyR4YfZEkTH~#i)|=L1YqHhV+H0M+j#`JU
zKexVX{nU1})7ofl-&{Ly?bG(KwZrj8K2KXGe4o>+o6mAhIA@%>zG?d%*Z-Z7zH7|`
z?TDUFIhx{or*!}p8$dbWjNii9O8vSjFuGdr(pX)smwaEOwx^8$UF#V~=U^=+Lfvb9
zruS#Qf2jO^!W?#5m#zP5?K1ubV;r`gFwO}_KLX`-YpFGN)BhZJr;Pm+th;INy!IGn
z^GlEJHr%_R>Iv|=X?4i2O@29!RM-KtPCc6Qf81K8b<phb&3qr<%vW5`czp*QyZPK?
zjBVQfjTuXsV|v<0>N>Y;u>YRE(()tM&v?J1O&46>U^Y8k`M&iZyr=nn2xWG-YU>T!
z%yV?XJ6g=~`xDo?z;A<*&Th_6(BeI3R=Hlkwvxt>wi`EN?1RNdqUAA|l-5eYSc12V
zM2d^l-f#VpHf3ICocYYPr>)-B0KEB>t{-ytn6r;>=CjN9d1~{dHI!P&U%8=V{GM~=
zCHzw`+U6ycU7(!P@-+4D`7%bcjdWK&oSWE`mYYD^LW*}2y)S^d4a_PASK#ODb`oxN
z$WL4S6<Ghf!r?A-tqZO{g&tB-i4v}`4jDrUkj@`(s9WaI!<FCB>WJBdp31%RG<xR|
z4uNw&{5lQp8)YM5Sf#`76WvC!J<oZ`32E{v%}Xt{t<s_#DA`YW`yF%>^9|0a@k;H-
z1Yev!LJf}}C+(<mNr~y=YVdA{UtP$9+^b@dAzw1C5+_#M$gg(kDetX>kz-2pyJr-z
zp3rL(SUZXC?Y*{u(sT2Y4yD&ISHh;*OBp?bi}FX!4jZGi*{d!|k7>1&biFQ9br~sb
z9@ivq3td*mLT8j4wIgg;jHN9J9Sz>Z4AnF3aPTN=QcuFKj1|XC`N=asWiBBtRa~D!
zpRQKl7mbugd+<)a>~TK6`)|GnZoUU^zK1&Ss-GF>mt)sQ>tno-7sst?c%9DTW$$98
zu9*EkS{PQyS*oA5dED~aCj63eVcWxEpE3FgFFj1yAtlXg`+RHR)JWw}ICbu1t?`PS
zs`Ewvth@+GnYC4>T~&LexpG=0E>rd)t@>-4cpI#N2ele}P^+PPTK$cAcCy7!lVl$f
z6>KM}sMFd;z0wXo^%M&1LM8i3@gY3Tf6!tT|Fl4Ka0S0-@m9M~`q|BE9*_7M&Q0>U
zdvk8=%hhMCSHxM*@VL4CyVeLJjnX5ZnQE;A`*~{(-?xMZ+(e?^GSW8R&(Za_z!>B8
zmXSxn<}D+RwO-I}jqAr8uX1$A7;E$&L_&<64vDGq2->()T0Ljf9$HNz(=VCP5nlch
zILq|ftHB>+=9}P^A;k9t6zGO7(~Q5zczsBg@;%P6c&+jKlz#6SWxqAd>w@+({2t@{
zTjC?nuj0p78Tn<x<BHk!aeM;JrQ;CTA91|GI0xYMnsNH+{gH2@x&z?7hlY>%p1`IL
zFpp#UozriMd5O;ykmabyXW(<pSVJ7g_(7*j;2i;d=*xBKxeA7FpvMS!&M?p2)K559
z3GNw`oB@aDyycU0pGkC`2bOf0;C%nh`3cTlF!nY5-}5~IUY`=)(r%7(lZ-c?;_yDM
zFEh$6-|wOOC%*T=<Lb-zJjXqpp9jtf5a(%k0_=I-&*(RkB(3`W9_s8O2h-ql0R7)X
z$7N(>k~y~1N4hWZUZl+nAa5smyn?>samjIide1@6EqF0U+tDPS*IfTZyA?(m<osvw
zxlZ3R>3jWV{C%)ChFhbRbY2IG^BelV2mem^PJ;J(nydI7-tfU_brZbC()d%r8sfdi
zIO4gEOvxvC*gnsYi>q7b8S-+G;4J|!<iz7Xpq?|Ye&lTr8jPoT%g1N53prPB)}Y2k
zqS5J%9*r@B6<*5!MxtvE*X83S@cN)f(6cI6(l_hpC~%~g<0vNs(7}Ga37_kZ^V|Z@
zCEBW&ZxTI@z*@U94V)qRUvpMGuNl9o|L>WHeo((Lg@ilaXCy~?9;cr=Je~NVJsM5p
z$fuzH3uuz<(p%=NEUQnUrxVEG1mmjbd*Gv8yXLpw@-_7I6?hl=@1jdnv=+YYK0`;p
zuYfz5sH=WFgQ72?^UE7NmBY}x_<l)yE%sod+k50heOjhpX&3S<Ka}HM<XYKOzU8y}
z5Z|+@-!2|Odm8r82&mvUjy_c9ZM%8;A3!(hE`RozuY5iP+Upy+&Gu{@JdHTkI6p+c
z*UY7lw(n^@!8!5L{!O3<Q)%8t)6Rc4;XOgS$#i|5?<Ly3pyd+%UNetZw0m_6enIzs
zcygZPM}6&#-@%>uz5EsMTTFZ%xv}S0VEr0whM30`_+2ypIyfHk)>d5K$nEY8eY5-z
zB|OzL`STq5tIv8)>7rg{yP)13LU;Yu2^@GuAKOa*d1POC+GSq)uQ$lWOU}Pe@}a$#
zAM(-bJ;0viTYoV{-xa8@B^ZZq?f8X%e-D0ZaPN?L9MiuueoJ@aC;z_{zf0_+`ZR(b
z=+CnMxXT`A`>lO2ZV9{mn!fU?3a{!{i{D)VYyH&({RiNmd@{QrpL(E~{JuzhkzZT9
zXNWHr@$s+Fqjfl#<CQUDa^-aidTXbI@&?=SnfBd$jxfdx&WzK-xa}psU(tJ$_e=V`
z0<yMtvh@M%JzfL;3{rOnv=3;{G^1aE(J+{4*~gH8DMs0+<rU+ZT{BWt8g?0LpSkq%
z9UgrN=rde-25krEJHQCTK%eH^81p#f%DaT2wy6hb+FdQ(1{g0UC{r9CFvd$-M7*|}
z@b=i4^N6c@UZaLRtkesx7Wybxujw~QWOcx`-SqlX>rdpux{$mR)+mwRGy7-OD6fND
zgT!jqm9}e&yb5`Y>^@}e?%K^>=Qk@fk6cUJ+t|IVr%Kkxm618CkIbj!FI@dqA$jz*
zkos$oJo;Kl12sq<eJ!NH8YGXt7Sd1+l1E<)srObM^&*!!Ggo+c`<nQ+>|GyR&2XwK
z^JZ2@?-+SN=TfqKv9d3#-M%|89{K{~K8&-p--dN{_EoH(GMkc5C|LF1fmO+H#(I4q
zwXM%GGMU0Fm=}GiW=YHS&b;`wXEyi?X4WY?<H|9=9x^{^bee2Q$%=izVwzLgN#`$+
z4s#kxiQn5d-}2Jqb8MV_qqVEeutaq}_NI8uJYvpjX_Y!<pN*IPP|3IC3}t(M_$gzX
z=`^Qqtmnw)e!9T_BN+X?a~0Lx#ANatIae_b7GwB_L40NQ8-M>?#WlX@U%85ZT&`jq
z-yFG$H^edKSd4KaSMizH)pc7x-`8BN<|<qZ=UhePE+SX)8J+t~jAE|B>qaKk97R6P
zvE#pT74@7|<eqNJ$^0u<;adKmF;`&@cOKnZ<I|j}YrRW!s}Jq?HFE~$Mkav$o^MC&
z!J?Z(SYaJChCZ&aE-_~?iLUpvLU>PGv)vP%8=&21?6JSiIk<*(?b{i@uge*<<2Umw
z;Fs5$c}@S@<_uiTnb&n~CcXO;J>Oti=UHo+1CE@5IT+Xb<D8%5<(mC7^zG;SfcJG;
zJ-z{k)f&HE&Y%;1)`EO1e*a84gAc6YccAuJT7^e^(<x_QPW%}fu?2mu_%@f>2j5(m
z55S=TR;6-pmh0x^K5>4Mx9fXX;?qeEo-xuMJY9s7?$qRX%C?8-pPBPzBEDCw+Q)gh
z2CuHZ9|PB@Ge@4)z3$i@t?rN4*9q<i<(JWPRDR?=%Cgevytip=?Pyk-L>+=_?yspt
zaP0aR)j9>Qb{ebq|BJD@*l&ux=D+b>T@^Iv@XKfD{%PmB4p$~k+PjbX@Ye39KD@R2
zs-0VZ`>fq-?F?JG#$Nv2ZDODQ?lw^aaCe)i54gKc)DGO;CRVQBCJT`Xi?wUyvhr{9
zPx-eQr2N|)Q2uQe=QORg@4`Z@a-D12ClwyZ(`HqP>)DduD68qLonEtF$tZNrI20Il
zYvT$!uZ-(BE9a$Nr(|~Ot?E@J&aAq!7qh1RmCBQ5b(KB4Zk0-P&uqc%EgbDW?94N?
zW0mVh#NQV3;ru~#Ep+dk$@wF&N*n9AqL$D)2Xh?qWh><+>U7E&Rp(=ETks(~PvJ=&
z@7whHq4N%}au<{;R!w_b$0fu1x+*rQSB2MJSr@)Gs)7nV+G%OFyj-ut%s6(ohEnaJ
z^R<S}%D|`a^=qp3@_dV)m#mq1++9QSJyPq=fL8MFQ$<s%Y~B(R8Bs;tG)|0sNyK$P
zq?{wEarQdKXyHDMahI<%;;Q&HN1Sz);xa2Cuf9ZJv$R-)dPZ)&^nb-{+*R&j?q)fx
z;&qJOw43DW$6aiz4x&oOmD|)!1R1+b$Lu%F)5ltbMZVp^8b&8aiJD$>W)-aG7~g$0
zvuT?^G+wlNU;&KXUvwX54NPYEwwlyxn>{|CfyFDXTleGM+-dqX2HYj=_9Sgqi4$jl
zyb2ump0)1_P|98EIj$em{|%9^bkg3gg24nBUN9^7?94o_Lx~}9egyXCe2#+sF!Wv^
zUfx2g-9L-pBg{waCc(?B-87PGRjhnl0QM|+x|?V1&NPwoC`ZfSzd~#4D%_`f4)*U8
z6gm6~_-9-jXLWi7CeJzd3<y_z_QR(Ea5)Bs)f?S>jsd$5TG~oXuE5>u2v=EWwB4hX
zJF-KJB1|hNjIWL97kKZ}&Yk!<e!Ctvs+ay3VDGm(bbgEL9M@NXYV<$Lao$gxg64ya
z<qBG^%rgIZz89dAyXNj{IuGj;PQlEY1@Z8BAG#>VO7;w`<*|FMCt!HZIODwM7}G4R
z6=L$*eaCU$?gv`$Fq_sKjy(VlR~);exJ&yZaCIeVrOGaFdJ-PfNY(<dUeJ3H_)~mK
zJ!_f9X=ROs*tpV{C+=YH@Hs+%<=-*f8K2;_46h#1>XaGv^X<O6wO7u=eeyYAn$uXP
zr4>1!X)y<;?&n$))(_N6+PPmXEu7mCGyeo$M?hL<Hs&K-742}&_~Zq@)EX-{W}v(k
z9kFf}Z`W!cIct5*G`N|`ctp=ZKCQkSp!Zwg4T9BY#vY~b8gRQgcL}UbMxUYW3>=;2
z-3>HrL!^-Vr>@h7q1!M$Pq{t_ymcTbpA+<O4|xnJyafIQE!+*Bfa0r(Dkrq+h9=uU
z9EGpeSv^lY?4hrC@6$poyBX&hu#eH!=Zt&}6l>5N;RG6vgJTbn)MhP~^?VDQUu5JF
zW;Vdcapyumxb}U4*Nsk}qYvtgd)Dp}s~I1Yl&Fc_VE+lYR%!Kd{tzB4@OuP&C&A=>
z;_@itZ}GdEvpbxTQr1Y!CLH(S$sA|ZtT7<X(tDM;y?`d_!aiEDMa#jYY4UgpEX@5H
zugxGWLr`^uS&wk#QJRYqu|glKFrR_f3|KF6G)db&Mj58Z9MpIJ{t-vJTp4GiD@HMg
za?Ga|W}0hO_qd+lpoM$j@^uP~dgy)3EBuOk_?OJfnmBdw49Lp0@_fo%!_QbJbV6@y
zHH{d@!0{y%bVprz)H?cm8BUq;dBr&Dn6+7Fd|QjPm5#M!+BEgmdIT$4l({Q-q!dN(
z-6Y3h?X)#Z%yEr&ON?z@PuSZrW~l60N%(=$t=ssBJ)Gpc)e2E1G|%iG(Po($>DiP5
z$G39BT*nR&tV$l?a~CYcXNzB2fj-`QP~n8*CEA}r*Ed`_L-wqR9E5HQKpBUtdyJs9
zkP@@Nu>QjuL@Q^O!ARPLC42_9+Mx|7{2WRy^Lvr2a|y%OKr~Oa1&l>nT01%d9CsqD
zSGR8SBQu`nIO=52fHuk4z0C3jpW0JB)O4bYw(|oxYN@Rn)WY=w^BKK9(EcN5miVQ1
zzE9`RIhsX6dcgb=I8kkAwT7A~b_2jzVtj4R5xuPKvO2|DNGp8T5=FG-`pQqdtjrut
z7NVEW6K2s1oKNY~SmF(5r#Pz(e8YPLSXys=x^iFzjJ9`(nQE<HaJ&os^d3t<+2fp5
zzw%1!X?==&d$aVg-xg=Jz-F(mprDmC2VC0&@;NP56XZ)~JxOotPOO8`U%pAy(Pj_R
z$E#}P05e`sa27b%i-fHtKE472>w@NhZjI6jBfQ|)>N_KwbB;EktoeNT-bXttbYB6%
z8Wuf<I5^%A6dR+*H5?eGot{gt{E=VAg4XBiF(y*GMUF3dy`b+SU>ldIC&p0=oONHy
zc+2rVaQ!u-=!JX1ZUqaVR?kAiceFg<*ITa5g3S_=6L+&IU;Fgk2kr;jt<mlf{FF2+
zxaWYi!0!+AF)O^CU>wr_EwogBN3ey)h8s{_|7b*I6<S}iM_MW?)VFCN7nKVA@)9lH
z(fbe|pu7%&$2;1o*GJ$o%9z?4_wn8!i@S`o%}n2b=`MXY;Q1jGw06lH=Q+pH%G&5_
zKDA~G(DR6^dz^a-3@b*b7~v4iju`I|{Wh83D&Mp8RTJLP)B4S8dd+fW0i0Gjqht*6
zYmm=V;2D9=^SKD7p4C>JbN&PHjF<f$q@VY-Sy8siNb}5hfd1Ak8MV7>-pl6!xO!c`
zJ5SGHMtu#AT9Nm(m;jz5KH}&SOsvCqC2`8Jb=%hVJI4*qnyuaEyqdQFwLfvjI?Wm8
zWX%7Pqjk=zS6W(Y?#%!XG2>6*WAxGsgf+03V+O+<e}b1*rdgA2WIoDS_g`XHReBuJ
zQ%=ZLqa&%VSF*1BjPusJTi35=e#_A;@?+d&lw)k}SgsJxX?0B-ZO9Rnu!>VkJ_7zq
zqNVrxec-r{>vR0}z7p>Z7)GWp8EKf&SJ9SZ=rInqb4liu;Un4&fqOq!^yk*r8DDP#
z&-l;y!aDCq$f<hU2QKEat>=G1Kkr7A7jb`_cJv*`n|#WpQRe;*NXA#2V50pUVJs!s
zsyuhgjMc3YcE{umTy%fRO1MQv+<*#uw7H<=H2CYGrZ_*zxN4)aG0o8iv|9wW_cQAA
zFPNp0V;1fmc+A7yUEYh#*R}N~wDb<hkBsKM32&i^d|m{)R%C;-Mjxj@TLzm?v{`2U
zvyAJ?PrT%>^qFE-J<Mr^w(gDZ0>3};@(mEh<_rwp^S#P@iay%?O=$8KidbVk!c`-*
z!z3w2)yvGywTZsk3VSQ4-tyah`ZZqe_qmq+2oLr+UZn3^a9RbLdg+?)6W6Rf{Y<Ml
z=oXwbT6LF1YpWf&0_G-HPwA=TX!E>#cnIi|v^`Dz)R+~nJcnmig;}wA4D{DvHk@eg
z?#l?jwb|YSbpd4SDOYIM$L#c|TD2qK8moWe`z5?{mA?%neV99Aj-(fu2D_Jh>r=15
zZ2^cz(pG0`leeIeYyU?;wPM)24c#?ar{y-7d<Hk;x&!(dZ>wGFTsfuZdcs>vtc~BM
zg)yR3GG3Lt!yJ8NgmbPgf}NGa7o3fIB+i+^8a-a~Hr5*j`a0)6CJpJKrC95q*17K5
zz`D|D-mbB?p~L_q8r2K{S3IQmJ7ji*;|Y%SFJ4&(BXKeU?%{id?*(A$O@_hYB{c4b
zg2qi+vH^Pb^WNc%k?10HRR$ixlVSR}bES81r9aDv-tqkiOfI3r2ES&Q@d)^O$K*W6
zz2Gtgb>Gq6d!3HCA|&Ct+H%F<YHI*YkD>B&&OQf2y@e|hV_fema2=(!RClMq;TcC~
z%v7DWrdxmfnj?LId+1}dKcnRw*X6N!1$W?#f%M|j%+hSnCOAD~Y$MuBDCS*7i_pi)
zacjx7H?BI3T6>aa*lq|&#_X;@Txn?az2`*CjHX=uPXkR~qG$1bBfVrFaJ;X@2)U2r
z88|%#6;A2-7D-zN({;vq&4_C7b<(C`KDDRb4W?W_r-j*>&y4#KTIn6M+&hV$m%!3b
zzM;h+SG`9?i|<NVFQHz40NQho&uI0Cc3L{`7aHJ-6}8qJyW$-H_B-IMGaqwBW@?`C
z%R2~_C8K^LTkk{Lf$H+bJ08^E_estM>7idYHn8g8BiEH_qjTr0Rvgn(TnFLFB6uCZ
zp*1iagDVT*;r%L;K-r<yBKWSscUK%{T;`aKap@quQV&-c#R}{ZMp$NE`tMO-Dxb4p
z=!(n?%{g2%5?KU_D*$b)t8yjMJ$UViUdgCwpB~<;^OiH8xN1bb!tWX8pyhWhb&%%c
zooGD?`z6LQ;?#b3bKTYL0i!PPdzqs{U|a*sYVcm>c1X)ty!DjhNjB%`u}Ew0^OaUt
zoIL=dBO2%W{DBodN4VP0{9P9f0@(_9xw}Ricap6;&^A72wik@&KDGAWjDb<BcP|;i
zJg3b7EhZV!`*)rL*Z6Lf*AY;yKOAJvSNw97ptg>Jkt_OFoHGaU8F?_T=m_RZTp50(
zm2rSMg!jPqUPJAud*xa^^MYCk?*=+$HuF$M>F%Soxh`|6qhMlO{)TU@wVLU=d@J#J
z5Nvm8Gs`t)?ImOG^4mzsOnAhu`g`YaiG*zMOWS9J>U~KEywx7pn#P@*ypMrx1fx%U
zN55(2cb24R7|hLmZ1Ao8YNc1`*_~P%tq%fWn?4(fXCL^yrkx&QnR)1k%u%>%8igKG
z*Olc5j*METf#CfOo6y=^k=f@FFg4?8{g)c8XECd#zdB7l-yqTRoP9|<<3Qs+*NrQ*
zIA^x{em#uwf|&`Q$Y6YBMroTF_R`Nxv6(6@hLN9EMJ?Lks!y!&*$Dl#_zYg&S+xy)
z%!%yKdXQh*^f@@7UbtG9+D6#Qkg?=4Z?jd~jQk0_)hw-BH@%Pfr3M=_7z>Ti!>38~
zb50L-1&)32&O39XLe>f+qyKBZFX^o%8>g2!Uo~BSXP!YXKM0>MkbPxI>u2m_L~RV?
ziou-tFuh;V$1}>Hx~~*pf`#(8kAzwwqJOvozkaS7^_v+VrKi?ynfBTOSFO&=J&`L$
zIRSTTxfVG#KPR@<ju;oX`dbHfZyM_(P`i0wFoN;D)mhpdt$oPW5OB<TEAu0?JZ41K
z_xkEn+AnkFm|sTci}W+jIARR*((Yv#XB`2}*hKDlALCojFY-DD#t7JG0SCd)*u5JF
z*Nn3WZM?H&l&kJn7*~x#Gb099isn{~!Hj90r!AMbW=_`ogpK{QaOY4|`8P*=4IO&<
zwFn2Kgil{kPxM<--nh7%*2dfu;JU?m{ZcPI2f@R*%b02yjKbrG6un58xQbCCT%VZz
zcc!juoZl6rXhG+hhj*HsaBYBd-3g~Y#_orMV;rA=-4G+WdQtYCGf!jFeyIM4Z|k95
zeY^5-<=#zSbJ$1Bz<9^ikZViV`vZ*if^YAge8+su@@p53-yd<!Jd`srXQVxlQ!Dg5
zNF%utFmhc<JT&rBUq_*Y(e?tk$zjJD;Jo*yeuh$V)K!;z7Ou^$%MvR+t}Amj$h(bP
z3w@-Aw$3Phi)-FrYP&^#Pk~D}d~&2dKD7iF{C<>1cby^62Ep}&c4pZgF>|HLoUK{Y
zX+{`@rmpUdBaeY<?p$e7Lflm_gM0{8^aaKf@$zmjSKF_F-jmKvGm3GcJ2+<J&25@{
zFv|3oI~K;PMv?j)e;Lu5q44DzAXi?w?`EX(4(h-Da@PHaWsco($ajqm!+rD0`X#fq
z-Z%Cdz8x^?G)Io@n$OvKcbQh&r<#n@%KfT6=3&m*n7WS{n4wUP-0@%K^8;;Oqm4do
zK#St@Zw|R`j_4JTp3&dfa}Z2FG275$??;zb;u@{AE$&mD(8`sCyAWDiIo6$|ev#HA
z{8EFLXk|vg+`tBXJ}?KP&12dck-6G%b!`l2Bx!}H89%LuQv3+G=8dvd-$zPj(kRAF
ztLe2xYu7DW1EpP0Wuz;{FKBK2CEr(p`w^VYMHxB!{GoH`y9@@-%5{@g+;xRg=AMAL
z8t;qMav8<C8(_q44*Ub>w9Ag;zT67^{M6sMqc{a+jPP9*8+ESHUwt+!ILGK(EVCo(
zkuiasGyAXq*7kgcNAlzX80K6eN;5i}qRl$cycf6^DR(Vq9#czfj>i2C?Zh+snT?kI
zJ_+j`SPVm%p@h#SV=JewK|H6g+kq?IhwhG{@z*10s_%C9y&GKI*Klvbr%=d8;cWr`
zjBB5v<qWULYH8!W*IJ+LnnxWsdtq)un`dm~3|;d^rppY}C-C-eN$u+@Bb(84T{c4x
z_fPfjJ<P)h&pR{+6V;3<dcgD%J-u&QKdZm;*;Hz#x@N}2d%BI0TroN)BQ>)qKF7g)
zwyW$bpjs;<HZS3dUd}yWpVPHUD|6vn{OV!sP4G9@;o9qh*4`!Ts!bYv=8Si<yQVOf
zbyTzOA$RUaoB`GRz8OU`TxNyM-pzAm8mvA5trtu~PORuPZ)$dM7CyaY1o^DAnML*v
zU@eU?!h62;!Ah=sE;GO~yAZhucPovnoRju@2n<~tPtju(+UeJfAKkgx1DdNyqx5N7
zn|ZQ&{4Hl*LYGZoxwCBs&-KL_GxFa1bw=!^^(>!eeZq#jKkDwNW4oR<8Zy$88t&aM
z0>KD7?5Otj_Pa;FzxDpnGw|~{WacNjc*hfptdwy*WPNMrJE5#nsaLcbUFMfp>-91j
zviIGYd5Sv(mGWcmX|30zRcpeURXl2N;5C3W#ILyT%X*$pH8Z&y?{;i&e;QYP4)tod
zJ|rl{{~fcq{aIe#0qGN=N)=D85BmCAqt5gEHL8xZT|$pL?V8jt&xtC}^a{GV!qXS~
zEV$;GT+viuhg3y1(+TpIpAVKR3A60^yeq5Y;<;d2nfwxFr%I^DH=_&nDSGBUsrMT;
zCDNKP&vl-mved-Zr=6Vx&k=-b2KEP}`54`bnY7<WSjXn~*e6#C@pkl1<4BvmM&04<
zb%n}grk}NeA(3~}O?t#r-J<r;b&S<&q5BV|lshBZ2W_gq?wU{3?EQ1#y2qSut9!Km
zv?&*!Q9f_$qiodeNS5+qw>rTLJUZ17-i<}49_={PZCSxYOSXF>&-Y7~yJ`8|Np5`Z
znCnoVESaBFV7q38bFpU`$|v;9U2h}uJ!ot1T%+97*|5X*RG+k}){2?^=t?uGQr{<$
zmf5;~Fo$IYw%H8hzQ5gzUw*#fJ9NhTCw(59m}<q2zvw|H*kSqe6-RoAP6#=_oaNor
z3#^RR>pFcu)^$2%y{_Kj(xiO&VsWbSD)Zd)awN~UN@brR8?^6&c3I}WZe|Y|BabG{
z<0-c8Py58ilO$o{T8|K1S?8Bm@5Zf3nH(FGyMyXj7R#%Th`Y<pQ;Xl-Bi+x(+vfK$
z4BBP%Tk*L~uCknDjqL1~s??DW51mEkcvZgYn8=%qgOI;<eD`eP4z7O;7t8p+9S@6q
zccxy!>2BVYcAYIllXJK8t3YZWvDn7)G}ilhRUO;s?_@6PmHNNLKKJWo8NF_q>k?gB
z6&wpq#I}l0!J%rWZ--d0|E2TJ@nIQJ`u?(6m(lI}5G;Q?NzRtwZfn#&1HFcxKccd@
zW2nv{b)-Byr|BuYS|XUTVtIVyA+>+;q-AsEuCk=vL*fRX5tZN3=a;Wgs^cj8?(gms
zUHKIDBKvYOp3hQ$f>DzWV$nW-&)zogfZ9G|Yee#sr#r0!Z+n7^ki_;$irKW*t}nxC
zi<?p7-x7VljaF*OL)K!|`@+hz)C*<BuJhAu%rwQSt6c}PEIZpKz00~S@+VE&K7`Bv
zqR_+jK)aB2!`iKf(eC!FTU8_5pTPcKX$_-}`hNkD-LqQZm?Kv@u2g@ZP2}CGQF|BJ
zyZn4AGaBE~GG30D|I?t_!z!n8r+QP@>#>zO=i7gzID-o1A1FLy{!fa=KV9VU^8a4=
zBs$kuf8zfL#9P{kkvVQ_V6+6^)3+C`?jnENxkZfAo5qO$)9|@;8LoBD#^f1Jz_QNC
zv(dNjOW#v|neTOlnR~_>$kl3OaI6>!&YzHD_AV`-ALZQ*o{u%LPwj9YL0X07G>_<t
z*|k8NjS2?o?sKs9xbOIk|6>q?+%9%#gzdUL_w_yi&snLfC-Hnpf&N^K%#T%ZcP*<u
z%(!b^H#w&zl5bg>m9d!9e!?i=#H?dB@r1%AHcF*W(6@@!(fyQa(Yij{CyDY>`@1AN
zaW{76nbLg%z5G9n;C^3{04saqd?&tH7aIT8psZ9m;>_K}&y+4$MY|I?W7bce2kl({
zyW^g@A})?BZ8MMajO1_3F>_#>wsJJ{CioQeG7}+g;$*%ta5cA+Wkb&q)w+(Z%#^tZ
zQ5gs+Q%3@xbaD6E%Ep+b8WWPN=M!e+O`Nejakc?PEX3To2gm#_Ja8WB<q>Cv;aFxC
z^=g@-Q8D7YZB?a~bI+8D)}D!xZ-eGNY0f<)so7h*Tcuw}ruU9o6)DBs;TD3n$2;cL
zCFwEGFSXMDyivpCrvLF`%Q%0XuH;$S!pwK9+^vTAH~z)K-6D4@{zzMS-=tp1f)%!2
z6I*HSU7y-Vac`G5YuWT6q4m;5URv*Q2Ifk8RadHZGdz&7fIO&L&Cr>}#Ixr0yKQEj
z;vv;$GP4F5tE%a8Q4MvZu&WaZi(=EVMpS!;1uGDmx?Iikg{O!-k}~NXQ+f+`*W8QD
z{~4qm)))HE3Px4uP|RNL!b(f;mUEBMYM0Q~(9!OOY<j<YndjrmJfFb9Q=okh2lX81
z36Ib)^};%&c1gE&cd~Lnp+(wXN4I7>yj$@$w?U9PoqLumkrU0fR7p|d<f3(5;Zt%X
zY<*%_YAIH=aY|fRq=(qEss>i&)ql>&*57}0Xs1cGshzB3mX{&8b$ic@U$L*n+Ua$i
zy|Q_N#A=Pgp9PiF_VBb}1GHDlt4FTDwHtcC_!lZU9rj&2E03DmSRw|gP%O2c(kg|p
zszQ9nOBl*&cx>%VSK2>xoa}|e7PvDM(Oux#X&&`?HfOQVPO@66K6E8cNVlkga3&>U
z$Rk8sDRehv@1?d1N$kDjEB;YO9?Yb3!53|va3V75WIOYYWcCMINO8^hv~%t@ymF8F
zoTa35&Tu1=$&s4&A^ych9}#<_851=wqA6{*)GBMqymFIH!NC~)sH1FHBjmqziHG!Z
zhR!h}yy|SLTBO&=oGX0oJj-I|<fvH5TXzbT_<C%ZAw+IGyiGg&Dt`lurwt$U+uCUF
z<j<09_CzU;Sgf;Eu1kS;L@1lpSyf@&Z=-)1ykeK-dDYgJD|L%qkA911jClI=i;vA7
zJ^S8#UAV$%dVpq}XS@?qcmZBT&+7hO2~v8>3Zld-YE(V0iYMSiOd9JZZJBE)Giu74
z7PcDShTVt=u;^K*2u!Z60U;tPbzeKGf5|%RxmXcs<szmHEZS{H)#`WhEuqU{GsP}b
zx7bN#ZUYYyBem;IMB!mCBG$|PY%@_Ke6W}rb>8ktv=h$E`Y)}h^pEeb<#m3EpZ_!%
z+U$r`?Jlu<)_vD}u1@N9BU`(T6bXk9cg$i@w2+ZID-?++z_GjVHr@+Shns1XXCVIq
z4<mBrw#2o+txajyvd*?-IZ@U~Gh6$13Ax{bJOoOy4qs<C!he)iS&q#6ll5ClQdh3t
zW;4u1D?f5H{D@I>RFSC>u8CX~SzV@GRz|9^V0Cp9v2;KR2<=)~u(n;uMyw&lL;oEA
zP^#4kJdDu7dWfIcS*@a^MNL7GPO&Y1JM`amfU#xdaI=0Ydx2ryGP;^pSB($UnQD$s
ziOBLDw5n=Dr`)v~ZQYY|(zdPeOH1UZGcR^7B9^K>`-!VY4btmfT6=is-4$?(9>;pe
zX!n5gj^cX7)j(bey#$Jptu3_j7yOpjO2Y;%!}4oQw6Km<ZPkndV(s~7*2KS~zjnon
zW~Ice&O8+)YY9`W)>Z50*8gG!_B~g;U)%ra^InWD&PL1cIO{#-W^t-#t^9B<>R;Bu
z|4A|^4A;WK{U6{;$&%d+Yn<2n(W;xaLiz7XyPOSu%#5uT4UcU<G1OK^{@%SFW#`k)
z_kSmPb%u0XpDucDvBOf86z|oP+Ir;9(ENu}?XtX5`i<$#g@(oN+{fAH_%?Xl8G5FE
zP$p<7zrr><|7uj>(+={uMPjpk2<@$U0CglsA?CDP1<6bKUucxIuxLk#LG)&>VfBaE
z>o?a8d9GPvixml<9(E#PAJ16%Y<9)9oiMM_mY8pzYdk43=ObOWshbhCMU9;tsb{If
z&OWcTHCD1BuiiRrZ^>AfHKhLq6FsDllvXP8yu<3%A!eIt99G3r3by<HVk^RWl`N#|
zTlwJm7^lQ$MuT3Lu0g-7hneH;cD?Eo?&F=35wS)@A<C<>ZpNJ<LDt)9m(&4UL{7Vj
zvpm<9w6D6G>ubGj<Uz6=MrJ}vc;~LT7{9q=phd2a8<BF>fYP&yz3btSHmMhw=G9${
zzQZik3$;6Ak^2<<qNmqOgx4;)eZa_)s|;0FuVwA3Ur=HrZ)bc`ZD+(Tu3Ep4*81vV
z1xb(x%1F++SEZxMeesVhb>JTvYAKoJTKx?f$kLT*AU*TBGUw2Zi^N-@C@*GjnzOCW
zqA9UO6WS@*U!kDbd%s^q{bqBl2~acDetA}<q~2)!OAlpb8_r6pX0M=Xf6cn8C&lwC
z!CAj<hoii-ueZcniHns=6{Fxm&KfJ#dXvb4=g|snnpQ%o@Gfbo89HNB7Tk4BeVyz_
zwjB1(nEP8!`kA8<*Fas|-%!ueM4T3xxQu5#yItjZdo&)n73tDmWDlr~J4;$tVhZ&;
zsFLx{bT^};=dnfXGKA(is|+W4=B$nsj3};Zy|aAg?4p{c)ar@O(hh;z6w#JkuUk_+
zX!t<kDASSIl%iFB7)yEQvDqf|CT0=X6deipIfD>930$>q%?MA5<eFQ3<4+u0)2Xig
z`nmX`Yn|6(RXL6<LUT67<E??I`xWj*=?6B@)BAGq-v(ce4thC{hva?q`)gk$Cb%o-
z)W<Awt|N?;gsDdBA(}0fO660%REp|5=v8fzUb{fj%19ISQJme+DEWi3?#S2m%i>fl
zM8uzdI&#FN5vK;VBa*1+T#Dvbaf&siaZa?((T1Z%RhGQY5r{srX|biFw$qV2@6tJP
zWDK=c5pCqTbh4jns`NaBI_6^4)uJU4jTTFtA+@j3@;=p5WW&8Ct*urs=d8Af>OFF$
zmalg+Z=(*+$|G0Xv76;x=wY$*ad>TUi+!!|R6jFP-ACPh)tV`vDHK%_3rS~=@nfrL
z=1t0JQDXN>iFei5<i{S7g*1!G4sA_DX$9N73aI){<Cf_!ZG-zts85!a7DqRcNZn_y
za4lpwV#myj62JU%ZslD?CFM!7)OAZzIHq0oT6YEWpP`QyzKXf?ynk-70&kZ`z0e9g
zD*9cIY^v)X^Om-&MqrsHcVo~wl92YsX4-`C-x-?gkdldSdaZw~Cn+-b1ZmdKYW<~b
z?Aul;Q6%Ml57D{buVeW;S^B0o(89_OBLrtvjjyy*Rj%tXwBM1l4-2S`cb#bLYWs*@
zteq&8OwliFsk$EC$S3v6<A4zQm~n`4lm~6BzA8MFPwdoI1g6<`g@4jdDc2Kx=V&^&
z)Y9Iz8Yw$zF)u0PI>mX4p_VUaCnF;zl|p9CIz&GECvddV%<P7}{>0m7g2g|NY$i)n
zS_EBlgjVevHmTY&;>4^YR-%Nh&)zLx$x<gqp<PwD?HJVIN~!SJM(@TNRU1)YM=!mn
zk)t$^h*}#F5Y+1?<$@z}J62y2=~wal0b2PlBHo+*BXx)7#60TbNNuQ6B(I^xx#~lM
z6GkLe$Yo5S3bP%ekTOrjGFAtXp@=9Z?h%nTO|5Y4;R?oRrs^fbqh&5-3fiiwQ`ypr
zwbCV3jarPon|sM}WF}!M&2zR$oN@#g+>s_Z`|PuFV;vv`!}CV$C%3oLY$A@4df6Z3
zYOe4Qw?`p=AHUWfX)Du_wU0ZNa-86KGw6|1k$TO@!AP$2da&$0<(N{7CIx<Z<QIIa
z#&~iwuxgL$!fsWoC4*yIjD0(;9~SY+Z=?07M9WcKwv;6ztmpRX@m)x6iQ)c}NFjDT
zl;YTP{1fML6jo&Wl+1nS^TsQ`MjTkpMU`>uvpM^cL5cnCIk1k*lW|==r>P`$+Uc(6
zMAejRA3{r-aD=QU)gv@5vRdLe%3@Tf%AZ*MI)^_a#qup@Ekhp5?CcxuTrcMyu@4dT
zXhv_^U!T2e-s)>0=rw!_n;LL;|L!DTc`o%@gOXEgQYc5}k=0(fp_02W&6c<EBvvj)
z=eKi5i_lKX>Q1wfj!$=W71WGCtLw>#WI}theJviQX`7p~s9X8SOLnSQigk0uGg)pr
z&8|p8)56tv77o~BQr5$-XjeQ6t&P3L_SJx#8A$&AC)yrybdcmVemCL8jzSfpG;d}M
zgNOAU3GJ%&fO$nxr*NKqXB_5hbBfB_Dh~3p9g8{_jO!xbVGLExe^fJIk>ASLG^yt}
z*<XZrj5WA*(m4+h(S#mdFKyf_m-Xk7OL4z7EK-S;@^~R{v70*s_neiv)POG(Th?*x
zgh4x=QK6)EJ9E8%b%!8B_IkGV#x!E!GlGurXAud->^spV=LBQ^;u-Nsl?qMijZ8~5
z`f`0}Udjxmkc_sgS;LZ(%<B;?vUnHD7>BvCFb>Wz;@*QMovRWl4cx^xcF+Gr=eelc
zv3|0hU*?w7O|N7fjdf5+M`xML(U~!swVQ`fSe&bK$e4EGZk1lfe>rydjvZHYxzf|;
zx;mCmV*L^5!H>H9nrV|(jve`-;s@lpRL)jV%NjCKU<6FDjd9!UgEY)>ZgY&t*w4UW
z^{P6r;z?r1x1Np3dU6~8hErr4_WCaVN!_TE@65p}Mx8uMXFX8zqw+b=TiUorE?!ou
z1nq)jp_BT<z&i8)TSjfCwz3&Jj1jXoB~Yv!xo%;qa+I;o5|E{*eTGe4uWIq%Klk76
z3B!{Wui@^y^4BS!WAD3tPV~;)iafZM&N?u;e}C@h+sDw$236J4hwc!})by;mHxw%>
zqswAxvxU`nTRRc5Q{9oy9#oI(RijXAevH<%!Qma<BT&n;)`w=tSlSvbS?pM5s~y?G
zPElf(<Dw_c7$l&p*AZ!`k!qnfKKBwgEt+~6E9q#Puh;cE^IYuoO2+KscH0bBk}PCR
z?-c*3HTsB%OSG?^35aF^l%%+CN6ihoht(+e|CC;No<fX%VO5%%SdSSUvD_VGG&q=|
z!KK>yP7$5+6WcljWgzFa>U*WqIA|7CV|p6rWn@^5bLul~dLdhvSS~z)T=tyU$j`WI
zG;==Jw^sR{?|01EvR~3WA}FO;s+4tIeJ!MZM>Mm|yj>j{$vtL=M*bKx$eqxLO!cUq
zsP=UK|L(rQnR-2iG%A$wnMfh8aaZDf*hE_;yHrt2yBrY;U73&kQ8k~Xo=efXAE{&G
zGmZZMgzQ0bq$FL8R!TKiP)OS<9mOZ>r7?}Z)e5kP5!)jzamd__epPR-%$mEie}*BQ
z#4Ou@Jl|~h0!n+{+Q{$;{hX(8ji*BM!usS$SPBLXVG+eJ*Qp%xJ1l{;wYT%u6Wgk8
z9U~GgGzmQOER<wz+1=7v*V$~f1B2#@JWFTfQ<i{T>_ye$XZ)M}J4xtTvK>;RE*%+0
zT{@Z&>zs8iYQFkkR=&+hG+S@SbL|tWD`lYEcbcCbCVZbDXdi!Fx8;3vtx?>=81}~~
zR?3(yj(9IPSB;^SA-OAdvFi~tR;oH{Pp-iz(RVe27@i}vF?^y>93{&`RA8$mUndb#
zH!1-9ZgRr0vV8O>Ex1cQLO%*LH9K@gN{96pVke1NNfKkLEctRWV%*S((Az4$#bVUQ
z3HxMpQjH2?jo2RD$W8aTf?rC!zl@5S6m>ON&Fs~sT+HO65l^-Z&RI@n=*1@}gC)Bi
zw#RcqsOJiA&u?|w1B~69A|J}5|5c`K60ew(9_qKBNnS%&ZFk&3q8>)&TRqnqx)lB(
z)~0p+mc!C3b~&=wDLUbGVdQ*B)Mm?_k`0WSs=|S~Wr%nuYCv31#$Dzii$3f955FNr
z>sevFUp=Rr=TPWu{N>r8W43Dg^zc<xY8t=Co%#`(3SUd6b{M0?daBeGG;OG!&2}Me
z(zr>ls>FNG?jPFhZbZLwI%iSzjpbhVn4x~f=VOFTQO6@H$r@Xv+y4s)oi)2|EVs;=
z9D&Bl=LlV|*2ZNjI&#Qd-MZumqPicL<*axv=N{QD>7*UVacWp@$FxmEwXvtFZCBo!
zIf94sU2N5)SOv?wA_-xKip4DQ-?Vb|^@FwVxt3k{=CZDnPZ&!6z<-m;*CRTQ7GdM-
zIpUQhcY4B{X%Y*w?e70&d6Obm!3Q+st+2_il$0v@l6hctTJX#GQEzYUZHy4oaX;0o
zxhU<V60F5>RAa*&YinJ~SkfZaA;sI|vF?|K_SJ>58T3)Q+r3oGTG`Q>IX5xLQ9#(E
zSQFIod}u7Kq(rUm7*w==Iqc$M8chi{1EGE&C#_X0x7s|Uvv@{};{U#VHqt!oLU=N9
zP^V%qENp>%k_Pf3VoNDx?3H=p6{REFC2MK#mV7;nI&S<koMe6c=1jY6JKCSIFu6~@
z%Sn#-FDChJC#~)!n4Cgk=Os1WHLU7^82i6J+HQaOZ549xLm9WHXPvL#k;50JcRxv=
zxMRl38qYqVople&OVFe~-Yj)BD@mqy5;o4-Xt%ys^_*j2gxx(O8y%}TciVns7u0#=
zyQ7R)rAe-O&s8)3_T~BKe@t_>j<I?!DryScIIgy+x37*_x6<<4T>oEMf30Dbqf7s(
zC)-5n_=H(RwkUA_0p2&~(41wSt<==kJ6`ZE?z0OS^!ecJuwrInAqRV)V0A7T^6$B(
zHx9Wqeltost57<(G|q{7&N?3BU(Ww<LvhFRDGYwcis&I@E7|4&ozr%Q@k^BYGqP6Q
zg_SBrLZ3jP9O3@Vc{SiOFqPVj;Y0ALdbaYrNMNQ*xmKskbi%@Vhq**wsk#UTg^ocV
zKeNu6O}2WopZ|Xd`VA&3&Zk*K^^g>eo}Fh=jc@BRnlaAb@-yPQyW44%uao}ep3Uzp
z^O+V|?(^{(TGE4_sKd>^&Eu%#GJ9-Amv<<JrLvkUD4d~olIJEg<=cP5)uOoqP`-?Q
z?#E-}%Xn4p7@z;ZODrEJ+<r`0T4{DaO#js~=FkbIGky%N!gA&hgIVQwzz!eo)Az!X
zYn}b4$8+$GSv>-}xcvAatg9nv$I5z7%TvADSt6S}DROrkzWRr}6R964GjT@tmj_jP
zc`t?gmsP$vyN7V4;L|?4a;+02{a21$?vibNquJX|68n4bHX{>U)8|IzSIM15e4?(l
zOIX&0Gwx?Db}#FR(LkL;XB{cxKJPXVs=N#sQ9IBMBk0|E^mZ&W{6ft-^m1PO?uOkC
zz3G>m68e`8bjvdNm*kaQ5!5LdR=J%j)$;{w&)(*r{)#TQ<DADYkw|-+zb0Ol^LoaJ
zj*KjtvsSi@^Ng#0=9TY+^S9L2OIG6?&#PC;oxiKss<IsUjqD-ri`c4AqG$C?Rfl5z
z`ZaK)XA^FrQ{HK4*NDMHqfGnKu4K%mQpW5r_^Z;UXxCq9qXo^BxvO0#tP0Qb?B&C0
z(p2+v{+p=T`whnT^Dp=XrN}j@a(WK8GR9Skma9$K2yMG>6g~9qqm-HCu>)$MLK(l%
z>M1aTT7{~&wJRfhJ%uuERl0l|HL4H`hQ>s(YoNqb*Ec`3X{NZ8o;g-B|Df&};TV%>
z-*=FKi0(SksH^qwwAO1^^)pLz?56G0LX0wgPttnG==U1f1!|Kff8kni|A7Ae9lb==
zOI5Y+am{v=nP$xLjNgOT?K2MFV}!2P?li*;uZl;RQB`JQHtlOW$5-dnuIdb;WgWsk
ldiNfhxVm#au1p)3xC(dIR{lqZ+ts(%uD=}T``7;5`hU&<i**11

literal 0
HcmV?d00001

diff --git a/report.md b/report.md
new file mode 100644
index 0000000..07c7cc3
--- /dev/null
+++ b/report.md
@@ -0,0 +1,210 @@
+# 1. Introduction
+
+## 1.1. Goal
+
+The goal with this project was to give a computer a drawing of either a cats or a dogs face and let it with high probability predict if it is a cat or a dog shown in that drawing.
+
+## 1.2. Scope
+
+First I was thinking that I would get lots of people to draw cat and dog faces for me which I later found out was far to time consuming. Therefore I had to change the scope from recognizing random peoples drawings to recognizing my own drawings, which is obviously easier. Nevertheless everything else didn't change that much, I just would get better results.
+
+# 2. Preparation
+
+## 2.1. Drawing and taking a photo
+
+![The raw drawings](https://jeena.net/images/2013/catdog/drawing-taking-photo.jpg)
+
+I drew eight A4 sheets of such cat and dog faces which resulted in 64 drawings of dog faces and 60 drawings of cat faces. Then I took pictures of them with my digital camera.
+
+There was a huge difference in quality between the pictures I took with my iPhone 4 cammera and the ones I took with my Nikon D5000. In fact I wasn't able to use the pictures I took with the iPhone because it was impossible to find straight lines in them.
+
+You can see here the result, one with the iPhone image as a source and the other with the Nikon one:
+
+![iPhone vs. Nikon sample](https://jeena.net/images/2013/catdog/iphone-sample.jpg)
+
+## 2.2. Photoshop
+
+I had to clean up the drawings so it would be easier for the algorithm later on to find everything. What I did was I opened the pictures of the drawings in Photoshop and played with the contrast and brightness.
+
+Then I cut out one drawing after another from the big image and saved it as a black and white PNG image without dither.
+
+![Steps in Photoshop](https://jeena.net/images/2013/catdog/photoshop.jpg)
+
+## 2.3. Resizing
+
+I wrote a small shellscript which would take all pictures resize them proportionally to a max width and height of 200 px and fill up the missibg borders with a white background color. To do that I used the [ImageMagick suite](www.imagemagick.org) software suite:
+
+<pre>#!/bin/sh
+
+NEW="new_$1"
+rm -rf $NEW
+mkdir $NEW
+
+for i in `ls -1 $1`
+do
+	convert $1/$i \
+		-adaptive-resize 200x200\> \
+		-size 200x200 xc:white +swap \
+		-gravity center \
+		-composite \
+		$NEW/$i
+done</pre>
+
+After that all the images had uniform size and colors so now I was able to compare them in a meaningfull way.
+
+# 3. Feature extraction
+
+The next step was to extract the features from the images which means to find something in the pictures which I was able to count and it would be unique enough to find a difference between cats and dogs but broad enough so all dogs would fall into one category and all cats to another.
+
+## 3.1. Straight lines
+
+The first thing which came to mind was counting and doing other stuff with straight lines in the image.
+
+### 3.1.1 Canny edge detector
+
+I used a the edge detector algorithm called Canny to preprocess the images which like its name says finds edges in images. Because of my preparation with Photoshop it was quite easy for it to find them. It is not easy to see that step with my drawings, here a picture how it looks like if you do that with a photo instead:
+
+![Canny on a photo from Wikipedia](https://jeena.net/images/2013/catdog/canny.jpg)
+
+What it does is basically it does noise reduction with a gausian filter and then finds the intentisty gradians of the image with help of some trigonometry.
+
+I didn't implement the algorithm myself, instead I used the [OpenCV implementation](http://docs.opencv.org/doc/tutorials/imgproc/imgtrans/canny_detector/canny_detector.html).
+
+### 3.1.2 Hough transform
+
+To find the lines I used the [Hough transform](https://en.wikipedia.org/wiki/Hough_transform) algorithm. The red lines are those which the Hough transform algorithm found in the example picture:
+
+![Hough lines](https://jeena.net/images/2013/catdog/hough.png)
+
+What it basically does is grouping edges, which can be imperfect, to object candidates by performing an explicit voting procedure. Detecting straight lines can be done by describing them as <code>y = mx + b</code> where <code>m</code> is the slope of the line and <code>b</code> is the intercept. The line is not represented by descrete points <code>(x1,y1)(x2,y2)</code> but instead as a <code>point(x,y)</code> in the parameter space, which makes detection of lines which are a bit off possible. In practice it is still more complicated, please read the [Wikipedia article](https://en.wikipedia.org/wiki/Hough_transform) about it.
+
+Because of lack of time I didn't implement it myself but used the probabilistic [OpenCV implementation](http://docs.opencv.org/modules/imgproc/doc/feature_detection.html?highlight=houghlinesp#houghlinesp).
+
+## 3.2. Lines features
+
+I extracted these features from the lines:
+
+- amount of lines
+- average length of lines
+- average angle of lines
+
+## 3.3. Other features
+
+I also extracted the amount of black pixels in the image to use it as a possible feature which wasn't using the extracted lines.
+
+# 4. _k_-nearest neighbor algorithm
+
+I chose to use the _k_-Nearest Neighbors algorithm which only locally looks at the neighbors of the document in a radius predefined by the user. It assumes that the document is of the same category as the highest number of neighbors within this radius.
+
In the following figure you can see that depending if the user choses k = 3, as showed by the solid line, the algorithm will conclude that the document in the center (green smiley) is of the type triangle because most of this three neighbors are triangles. If on the other hand the user choses k = 7, as showed by the dotted line, then the amount of neighbors which are rectangles is greater as the amount of neighbors which are triangles, so it concludes that the smiley is of type rectangle.
+
+![k-Nearest Neighbours as a graphic](https://jeena.net/images/2013/catdog/k-nearest-neighbours.png)
+
+In the picture above you see how it would look with two dimensions. I have been using four features so the algorithm had to check the distance to the neighbours in four dimensions. This isn't really more difficult, it is just more to calculate.
+
+# 5. Results
+
+The results were quite encouraging, I assume it is because I only used one style to draw the dogs and one style to draw the cats.
+
+## 5.1. k-fold Cross-validation
+
+I used 10 fold cross-validation for every test I did, which means that I used 90% of the available data for the learning algorithms and then the remaining 10% to test how they performed. I repeated this ten times until all data has been used for testing once.
+
+## 5.2. Results with all features
+
+When I used all of the features and three nearest neighbours I got amazing 100% accuracy, which was kind of suspect because that normally means that you most probably did something wrong.
+
+## 5.3. Results with a reduced feature set
+
+Therefor I tried to reduce the features to check if it would perform worse.
+
+1. When I removed the information about the amount of black pixels basically nothing happened.
+2. When I removed the information about the amount of lines and average length at least I got a couple of wrong categorized images, the accuracy went down to 95%.
+3. When I removed the information about the average angle of the lines, that was when I got significant errors. The accuracy dropped down to about 60%, which is still better then pure chanse.
+
+So it seems like the best feature to detect cat and dog face drawings done by me was the average angle of the straight lines in the image.
+
+# 6. Future study
+
+The most important next step would be to gather many more drawings done by other people who use other styles to draw cat and dog faces.
+
+Then it would be interesting to use other learning algorithms like Bayes, Perceptron, etc.
+
+And then it would be interesting to use this approach on photos of real cats and dogs.
+
+# 7. Code
+
+	#!/usr/bin/env python
+
+	import cv2, cv, sys, math, os, numpy
+	from scipy.spatial import KDTree
+
+	def extractFeatures(label):
+
+		directory = "img/" + label + "/"
+
+		features = []
+
+		for fn in os.listdir(directory):
+
+			img = cv2.imread(directory + fn, 0)
+
+			# find edges
+			canny = cv2.Canny(img, 50, 100)
+
+			# find colored
+			black_pixels = numpy.count_nonzero(img)
+
+			# find lines lines
+			lines = cv2.HoughLinesP(canny, 1, math.pi/360, 5, None, 10, 1)
+
+			lengths = []
+			angles = []
+			try:
+				for line in lines[0]:
+					x1, y1, x2, y2 = line
+
+					# Pythagoras
+					a2 = math.pow((x1-x2), 2)
+					b2 = math.pow((y1-y2), 2)
+					length = int(math.sqrt(a2 + b2))
+					lengths.append(length)
+
+					angle = int(math.degrees(math.atan((y1-y2) / (x1-x2))))
+					angles.append(angle)
+			except:
+				pass
+
+			# print out everything
+			lines_count = len(lengths)
+			mid_length = sum(lengths) / lines_count
+			mid_angle = sum(angles) / lines_count
+
+			features.append([
+				[lines_count, mid_length, mid_angle, black_pixels],
+				label
+			])
+
+		return features
+
+
+	if __name__ == "__main__":
+		cats = extractFeatures("cat")
+		dogs = extractFeatures("dog")
+
+		test_count = 5
+
+		test_data = dogs[:test_count] + cats[:test_count] 
+		test_labels = map(lambda a: a[1], test_data)
+		test_features = map(lambda a: a[0], test_data)
+
+		data = cats[test_count:] + dogs[test_count:]
+		labels = map(lambda a: a[1], data)
+		features = map(lambda a: a[0], data)
+
+		tree = KDTree(features)
+		
+		for t in xrange(0, test_count * 2):
+			d, i = tree.query(test_features[t], k=3)
+			print "-"
+			for j in xrange(0, len(i)):
+				print test_labels[t] + " is a " + labels[i[j]]
\ No newline at end of file