机器视觉维修系统英文文献和中文翻译(8)

菜单

A common artifact that is observed with figure images is that objects of interest will often be illuminated in a way that causes a . curve of background-matching intensities along the interior of their projected area. This intensity gradient artifact also occurs at regional borders within an inpidual object of interest. The result of figure image segmentation when this artifact occurs is a pair of segments whose bounding boxes overlap (see Fig. 2). Since these curves are usually not parallel to an axis for their entire length, bounding box pairs caused by this phenomenon almost always overlap. Thus, our system merges overlapping figure segments into a single segment. This merging of figure segments can sometimes have the additional benefit

of overcoming the occlusion of an object of interest by a small obstruction (see Fig. 3).

2.5. Segmentation optimizations

2.5. 1. Segmentation subsampling. As stated thus far, the figure/ground framework provides a relatively fast form of object detection. However, the detection problem’s critical real-time nature is such that we would still like to identify ways in which the detection performance can be further improved. Such an improvement is possible through the use of subsampling. Because the segmentation algorithm is currently implemented in software, each figure pixel must be considered sequentially. The subsampling optimization improves performance by considering

(a) (b)

(a) Two intersecting figure segments before they are merged

(b) The figure segment that results from their merging

Fig. 2. Merging of figure segments.

(a) (b)

(a) An occluding object (a strip of paper)

is placed over a target (a knife).

(b) Ground, current, and figure images are maintained as before.

(c) The two ends of the knife produce bounding boxes that overlap. As a result, a single figure segment is pro- duced for the target despite the occlusion.

Fig. 3. Overcoming an occlusion through figure segment merging.

only every few pixels instead of every inpidual between these two share that same state. As a pixel. We make the assumption that if a pixel has the result, this optimization causes object detection to same figureJground state as the previously considered fail when the objects approach their minimally pixel (in either dimensions), then all the pixels acceptable height or width. The failure is due to the

fact that they are being pruned away by the • Figure segmentation is only applied to areas where segmentation process (see Section 2.4) in situations it is expected to provide results (i.e. in intrisically when the consideration of each inpidual pixel active areas or in areas currently near an object of would have resulted in an acceptable height or