image formation by lenses 
Why is the image formed by an object as described in the picture can be derived only from these three ray of lights?Every point on the object has multiple ray of lights going in every direction, how do we know that the other ray of lights(other then these three) don't converge at some other points, or alternatively create a big splotch of light the would hide the image?  
 A: With an irregular piece of glass it happens exactly as you describe. Three rays may converge onto one point, but others may go anywhere else.
The convergence onto one point for all rays is only true for perfect lenses. ('Perfect' being defined by exactly that metric - every ray from one object point converging on one image point, mostly with the additional requirement that all image points for a flat vertical object are also flat and vertical.
What makes these three rays of light unique is that they are very easy to draw.
A: 
Why is the image formed by an object as described in the picture can
  be derived only from these three ray of lights?

Actually two rays are enough. But no real lens (or system of lenses)
will behave that way. I wonder if your teacher or textbook didn't tell
you that this is an approximation, called after Gauss, good under
special conditions. There is a prerequisite: that the optical system
has an axis of symmetry (the dashed line in your figure). This is known
as a centred optical system. In order Gauss' approximation holds
good it is required


*

*That source and all rays are near optical axis.

*That all rays are at small angles wrt optical axis.


Under these hypotheses it can be shown that all rays starting from a
point (the source) converge to another point (the image). It may also
happen that rays exit the system diverging; in this case they will all
meet if prolonged backwards (virtual image).
The construction you are showing uses some other properties of Gauss'
approximation:


*

*All rays entering the lens parallel to optical axis converge to a
point, named rear focus.

*There is a point (front focus) such that all rays passing through
it before entering the lens exit parallel to optical axis.

*For a thin lens rays passing through its center are not deflected.
For a general Gaussian system this property is expressed in a more
complicated way, involving the nodal points.


All this given, for any point source you can find its image (for a
thin lens) using two rays among the following three


*

*a ray starting parallel to optical axis

*a ray passing through front focus

*a ray passing through lens' centre.

