As [Paul has pointed out][1], there are many different ways of defining atomic radii.
[ron has also pointed out][2] that the trend goes in the *opposite* direction from what you said.
For instance, according to [Ptable][3],

    |Calculated radii |
    |  |r(pm)|  |r(pm)|
    |H | 53  |He| 31  |
    |F | 42  |Ne| 38  |
    |Cl| 79  |Ar| 71  |
    |Br| 94  |Kr| 88  |

I have added hydrogen here as the natural "halogen" for helium.
Empirical and covalent also follow the same trend as above.
But the van der Waals radii, as shown below, do not,

    |    vdW radii    |
    |  |r(pm)|  |r(pm)|
    |H | 120 |He| 140 |
    |F | 147 |Ne| 154 |
    |Cl| 175 |Ar| 188 |
    |Br| 185 |Kr| 202 |

I am assuming this is the data set you've mentioned.
The reason for the disagreement is that electron clouds may relax or contract depending on their chemical environments and so do atomic radii.
Thus, what this shows is that:

1. H, F, Cl and Br are **larger** than He, Ne, Ar and Kr when **bonded**.
2. H, F, Cl and Br are **smaller** than He, Ne, Ar and Kr when **not bonded**;

The reason for the above is the following:

1. For covalent radii ([source][4]),
  > Atomic size gradually decreases from left to right across a period of elements.
This is because, within a period or family of elements, all electrons are added to the same shell.
However, at the same time, protons are being added to the nucleus, making it more positively charged.
The effect of increasing proton number is greater than that of the increasing electron number; therefore, there is a greater nuclear attraction.
This means that the nucleus attracts the electrons more strongly, pulling the atom's shell closer to the nucleus.
The valence electrons are held closer towards the nucleus of the atom.
As a result, the atomic radius decreases.

  What is being described is also called [shielding effect][5].

2. For van der Waals radii ([source][6]),
  > [While] van der Waals radius is used to define half of the distance between the closest approach of two non-bonded atoms of a given element.

  > [...]

  > [Based on the ideal gas law above], van der Waals equation takes the molecular size and the molecular force into account. As a result, the attractive force (a/V2) is added into the pressure part. Likewise, the volume is subtracted by the molecular volume (b), which is determined by the van der Waals radius.

  Here the shielding effect is much less important.
  van der Waals radii are actually used to express how much volume each atom occupies on its own.
  It is a minimum interatomic contact distance, less than that the atoms are somehow bonded.
  And here the most important effect is plain electrostatic repulsion between atoms.

  Thus it seems reasonable to me that it is harder to put two Kr atoms close together than two Br atoms, simply because Kr has more electrons and the electrostatic repulsion will be stronger between them.

  [1]: http://chemistry.stackexchange.com/a/66588/40029
  [2]: https://chemistry.stackexchange.com/questions/66587/why-does-being-bonded-decrease-atomic-size#comment114980_66587
  [3]: http://www.ptable.com/
  [4]: http://chem.libretexts.org/Core/Inorganic_Chemistry/Descriptive_Chemistry/Periodic_Trends_of_Elemental_Properties/Periodic_Trends#Atomic_Radius_Trends
  [5]: https://en.wikipedia.org/wiki/Shielding_effect
  [6]: http://chem.libretexts.org/Core/Physical_and_Theoretical_Chemistry/Chemical_Bonding/General_Principles_of_Chemical_Bonding/Covalent_Bond_Distance,_Radius_and_van_der_Waals_Radius