I am using this code simulate real camera to capture 3D object:
patch(ax, Object3D, 'FaceColor', 'flat');
camva(ax, rad2deg(VerticalAOV)); % Set the camera field of view
camup(ax, [0 -1 0]);
campos(ax, [CameraPosition.X CameraPosition.Y CameraPosition.Z]); % Put the camera at the origin
camtarget(ax, [CameraPosition.X CameraPosition.Y CameraPosition.Z] + [0 0 1]); % The camera looks along the +Z axis
camproj(ax,'perspective');
axis image;
axis off;
WidthResolution = SensorWidthResolution/(ax.Position(3));
Image = export_fig('temp.jpg',ax, sprintf('-r%f', WidthResolution),'-nocrop');
[ImageHeight, ImageWidth, Channels] = size(Image);
Image = imcrop(Image,[(ImageWidth-Width)/2, (ImageHeight-Height)/2, Width-1, Height-1]);
The problems are:
I used VerticalAOV as input for camva. I chose VerticalAOV based on this image. However, this is illogical. To get real view, one should provide both Horizontal and Vertical Angle.
[ImageHeight, ImageWidth, Channels] = size(Image); returns wrong values. ImageWidth and ImageHeight does not provide the correct values. It is clear that one of the dimessions would wrong since only one dimension of the AOV was provided. However, why both of them are not correct?
The crop I made is to correct the dimensions of the image. However, it seems to be not useful step (I checked the result using ground truth data I have and they did not match).
P.S. export_fig can be found here: https://github.com/altmany/export_fig