3D technology increases the visual quality as compared to 2D technology.DIBR (Depth Image Based Rendering) technique is used to convert 2D to 3D using a single color image and it’s corresponding depth map. But the main problem with DIBR is that the generated 3D quality is of low quality due to the presence of disocclusions in the depth map. This work proposes an adaptive multimode filtering of depth map to improve the quality of 3D image generated by DIBR.The proposed method uses a Fast Parallax Lookup table to identify the hole locations .Then the depth map is divided in to sub-blocks, which is further subjected to classification to categorize the holes in to bigger and smaller holes. After that multi-mode filtering is applied according to the hole size to optimize the depth map. The experimental result shows that the proposed method results in high PSNR value as compared to existing methods.