Image acquisition

The images were taken with an Olympus C-3030ZOOM digital camera at a resolution of 1280x960 pixels. Many of the full-size images contain not only the sample, but also some background.

A single light source (standard desk lamp with a 60W tungsten light bulb) was used.

Images were taken at 9 different scales spanning two octaves. At the central scale the distance between the camera and the target was 28cm, this was selected to correspond roughly to the default scale in the CUReT database. The scales used are described in full in Table 2, and full-resolution images from one material (Cracker B) are shown in Figure 2.

Table 2: The scales present in the KTH-TIPS database.

Scale number	Relative scale	Distance to camera (cm)
1	2-1.00=0.500	14.00
2	2-0.75=0.595	16.65
3	2-0.50=0.707	19.80
4	2-0.25=0.841	23.55
5	20=1.000	28.00
6	2+0.25=1.189	33.30
7	2+0.50=1.414	39.60
8	2+0.75=1.682	47.09
9	2+1.00=2.000	56.00

Figure 2: Full-size images depicting the variation of scale present in the KTH-TIPS database.

Scale 1	Scale 2	Scale 3
Scale 4	Scale 5	Scale 6
Scale 7	Scale 8	Scale 9

At each scale 9 images were taken in a combination of three poses (frontal, rotated 22.5^o left and rotated 22.5^o right) and three illumination conditions (from the front, from the side at roughly 45^o and from the top at roughly 45^o. The adopted labeling scheme is shown in Table 3 and sample images in Figure 3.

Table 3: The labeling of images within each scale in the KTH-TIPS database.

Image number	Object pose			Illumination direction
	Frontal	22.5o left	22.5o right	Frontal	45o from top	45o from side
1	x			x
2	x				x
3	x					x
4		x		x
5		x			x
6		x				x
7			x	x
8			x		x
9			x			x

Figure 3: The variation of pose and illumination present in the KTH-TIPS database. In each row the pose is constant, whereas in each column the illumination is the same (frontal, side, or top illumination).

Image #1	Image #2	Image #3
Image #4	Image #5	Image #6
Image #7	Image #8	Image #9