CONSTRUCCIÓN DEL SISTEMA DE MEDIDA DE CONTENIDO DE

4.2

3D acoustic cam era im age representation

Two-dimensional acoustic intensity images corresponding to a number of frame dis­ tances are formed (figure 4.2) and only the maximum intensity along each beam and the corresponding range is recorded, as it is assumed th at the maximum inten­ sity along a beam corresponds to the acoustic return from the closest object in this particular viewing direction. The data is therefore a set of sparse data points in 3D space with associated acoustic intensities rather than a packed volume. Recording of the points corresponding to the maximum intensities along each beam reduces the amount of data to be stored and processed but of course might have the effect th at the acoustic returns from objects with weak acoustic refiectivity are overshad­ owed by the sidelobes of nearby objects with strong acoustic returns. For debugging purposes the full data set of all intensities along each beam can be stored. The full data set from the debugging mode would for example only be required in case of hardware problems or when the reduced data set does not show the expected results so th at potential problems with the image generation can be identified.

Figure 4.2: EchoScope acoustic camera and two-dimensional acoustic intensity im­ ages

The data of 3D acoustic images can be best represented as

• intensity and range images, or

• points in three-dimensional space (voxels)

Examples of acoustic images of a real underwater scene captured by the EchoScope 3D acoustic camera in Fjord waters in Norway can be seen in figures 4.3 and 4.4. In these images the visible scene is composed of three cylinders and the seafioor as shown in the video image of the scene in figure 4.5. Figure 4.3 shows the voxel representation of the data. The grey level of each point is determined by the corresponding acoustic intensity and the viewing volume boundaries are represented by the solid lines. This representation of the data is the most useful for human interpretation and as such would already be very useful for ROV operators as an additional navigational tool. Figure 4.4 shows the intensity and range images of the same scene. Dark pixels in the intensity image correspond to returns with high

CHAPTER 4. 3D ACOUSTIC CAMERA

acoustic intensity and dark pixels in the range image correspond to points in a closer distance. Each pixel of the 64x64 intensity and range image corresponds to a distinct beam-direction. Knowing the viewing angle corresponding to each pixel, the Cartesian co-ordinates of the image points can be calculated to arrive at the voxel representation. In the EchoScope case the voxel positions for a given range image R {u ,v ) with u ,v = 1 , .. . ,64 and viewing volume angle 7 are defined as

Xuv = R {u ,v ) sinOy, (4.1)

Vuv = H(u,i;}sin/?u, (4.2)

Zyy = - R ( u , v ) COS ay COS Pu, (4.3)

where

(4.4)

A = + (4 5)

The intensity and range representation scheme is not easy to interpret by a hu­ man for navigational purposes, but allows a very useful representation for machine based processing of the images as it, although of a lower resolution, somewhat re­ sembles the representation of conventional optical or range images. Computer vision methods such as segmentation and reconstruction techniques based on threshold­ ing, Markov random fields [MTR98], clustering [GDM96], connected components [AM96], or surface fitting [SB93], can thus be implemented.

500 -500 -500 500 C am era co-ordinate system <1000 -800 -600 -400 -200

Figure 4.3: Voxel representation of a real 3D acoustic image

a)

Figure 4.4: Real 3D acoustic image - a) intensity image and b) range image

CHAPTER 4. 3D ACOUSTIC CAMERA

To summarise, the 3D acoustic camera data can be generally characterised as a noisy range image with associated acoustic intensities. The data can also be represented as an ordered sparse 3D point set with associated confidence values (acoustic in­ tensities) whereby it should be noted th at the positional uncertainty increases from near to far as every data point is associated with a spherical cell in the viewing volume, the size of which is dependent on the range. The volume of the spherical cell as seen in figure 4.6 is given as

i2

^spherical cell ~ ^ [^2 “ ^i] • (4-6) Re-writing this equation using Ad = c?2 ~ di and substituting di by d — Ad/2 and d2 by d -I- A d/2 results in

^spherical cell (4,7)

where d is the distance to the center of the spherical cell. Ad is the distance reso­ lution and (f) is the beam -to-beam angular resolution. The distance between two points with range distance d and angular separation 0 is given as

d^ = 2d sin ^ (4,8)

The distance resolution and beam -to-beam angular resolution are constant pa­ rameters of the acoustic camera. Figure 4.7 shows the spherical cell volume, and figure 4.8 shows the lateral distance, as a function of the distance to the spheri­ cal cell for a 64x64 EchoScope image with some typical viewing volume angle and distance resolution parameters.

Figure 4.6: Spherical cell in the viewing volume for beam -to-beam angular resolu­ tion 0 and bounding frame distances di and d2

900 A d = 5cm, Y= 25deg Ad = 10cm, Y= 25deg — A d = 6cm, Y= SOdeg A d = 10cm, Y= SOdeg 800 700 600 i 500 1 400 > 300 200 100 200 400 600 d [cm] 800 1000 1200

Figure 4.7: Volume of the spherical cell fgphencal cell range distance d for typical EchoScope parameters for the distance resolution Ad and the viewing volume 7

CHAPTER 4. 3D ACOUSTIC CAMERA