Detect edges of object in relation of the player point of view

Hi guys,

My player has a circular type of view (black circle and he is on the center) and i need to know the region behind objects. The problem is that i am stuck in how can i identify this edges (blue points hiting the red box in the picture) in order to create the area behind the object (the area with black borders). I am running away of raycasting until hits because it will lose performance. I think the edges i am looking for are the mesh vertices that hits the bounding box, but i dont know how can i have this points.

The red boxs are the bounds of the object.

Sorry for my english, ty for your patience.

13603-question.jpg

Here is one reference to a number of different algorithms for this functionality:

http://negativesum.net/Members/hoss/resume/SED_Algorithms.pdf

Another solution is to break the object into scan widths, and then walk the vertices projecting them onto a plane. The vertices with min and max are assigned to that scan width. Connecting the dots produces an outline. You’d have to do both a vertical and horizontal scan to get all sides of the silhouette. The smaller the resolution, the more accurate the edges. The solutions in the paper at the link provided are more accurate, but scanning like this gives the possibility of create a vector outline (though I’m sure sure how to connect the horizontal and vertical scans), which I don’t see how you could do with the algorithms in the paper.