Java components / classes should be as independent as possible of other Java classes. This increases the possibility to reuse these classes and to test them independently of other classes(Unit Testing). To decouple Java components from other Java components the dependency to a certain other class should get injected into them rather that the class itself creates / finds this object.
A class A has a dependency to class B if class uses class B as a variable.
If dependency injection is used then the class B is given to class A via
the constructor of the class A - this is then called construction injection
a setter - this is then called setter injection
The general concept between dependency injection is called Inversion of Control. A class >should not configure itself but should be configured from outside.
A design based on independent classes / components increases the re-usability and >possibility to test the software. For example if a class A expects a Dao (Data Access >object) for receiving the data from a database you can easily create another test object >which mocks the database connection and inject this object into A to test A without having >an actual database connection.
A software design based on dependency injection is possible with standard Java.
Spring just add some simplifications in using dependency injection by providing a standard >way of providing the configuration and by managing the reference to the created objects.
For more read this
Edit1:
When spring initializes its context it creates all the beans defined eager in spring application context.xml file
Now suppose your Bean A has dependency of B then the Obj of B is already with Spring as it has been created successfully while initializing it will search for setter method in class A and will set B's Obj there.
Edit2:
Please read 3.3.1.2. Setter Injection