%PDF-1.4 % 1 0 obj << /S /GoTo /D (section.1) >> endobj 4 0 obj (Introduction) endobj 5 0 obj << /S /GoTo /D (section.2) >> endobj 8 0 obj (Related works) endobj 9 0 obj << /S /GoTo /D (section.3) >> endobj 12 0 obj (The NBNL approach) endobj 13 0 obj << /S /GoTo /D (subsection.3.1) >> endobj 16 0 obj (The NBNN algorithm) endobj 17 0 obj << /S /GoTo /D (subsection.3.2) >> endobj 20 0 obj (The NBNL decision rule) endobj 21 0 obj << /S /GoTo /D (subsection.3.3) >> endobj 24 0 obj (Learning the NBNL prototypes) endobj 25 0 obj << /S /GoTo /D (section.4) >> endobj 28 0 obj (Experiments) endobj 29 0 obj << /S /GoTo /D (subsection.4.1) >> endobj 32 0 obj (Single-scale experiments) endobj 33 0 obj << /S /GoTo /D (subsection.4.2) >> endobj 36 0 obj (Multi-scale experiments) endobj 37 0 obj << /S /GoTo /D (section.5) >> endobj 40 0 obj (Conclusions) endobj 41 0 obj << /S /GoTo /D (section*.2) >> endobj 44 0 obj (References) endobj 45 0 obj << /S /GoTo /D [46 0 R /Fit ] >> endobj 88 0 obj << /Length 4682 /Filter /FlateDecode >> stream xڵ[K8rϯQh㋷=v9z[Ӈ:(Hb4EjHk~L!z} DB |'+w?Wǟ$Zynzܯ UgNyiPm]͡.ZG>yZ?oѦVemW^9n~Y,|ѺmyEimml<'u#-_lM]iF<;$V[w$2?hBf#}Z/5Vx4m;:Q,~&|,JOoM֛X\5&f(^8t( !#ke86;ݍ`>.k''Y&p^KCǸp~;篂ڇ}\~|hF<·\OV.o) I> ~>v9座>dV;i/rrN6;AUJ|:O2ħ*y6/z=l2!")0?-뢽e^Igoy)=!4 ԩ!P觪CUނ Y%Es#em@RVe&!n~qj`д60*#Ǧ3wxEQaa>^Ny-S3iZU^2$bt2]Eyot0*5ۏ/M#$^iϲԽ riᨔ 4F|gIB&fUu2 miO8c)vuB%}HkfhͽDK[0p0&w6g⬜y@-A6eN0eZtݕKj{#r
H ɩ1
Uz= Ε$<4kdL8bv"t(k|zsFpB/¹j<7t/K\=::i;\n}yjLj
ј<