Results

Frequency Orientation Test

The main purpose of using this test image is to see how well the algorithms remove alising in the red and blue channels.

Figure 1. Original Frequency Orientation Image

Figure 2. Demosaicked Frequency Orientation Imge using Bilinear Interpolation

Figure 3. Demosaicked Frequency Orientation Imge using PCSD

Figure 4. Demosaicked Frequency Orientation Imge using Alternating Projections

The red, blue and yellow coloration at the high frequency regions of the demosaicked images are caused by red and blue channels aliasing. The main observation from this experiment is that Alternating Projections is very sucessful in removing red blue channel aliasing. Demosaicking using Bilinear interpolation creates most severe red and blue channel aliasing. This shows that red and blue channel aliasing can be removed by exploiting the correlations between the channels.