High-resolution images are needed for accurate pattern recognition applications. The images captured by cameras residing in an aerial vehicle may not be able provide high resolution information in the specific object regions on the ground or sea. Hence a super-resolution reconstruction of the images from a sequence of low-resolution images is required. Super-resolution image analysis uses information from several low-resolution images in order to compose one high-resolution image. It involves two important image processing techniques namely registration and interpolation. Registration deals with the issue of aligning the low-resolution images. Though the original images were taken from approximately the same location, they are likely to be subjected to slightly different tilt, pan, rotation, and zoom. These images are registered or aligned appropriately before the reconstruction of the high-resolution image. Interpolation is the process of comparing the low-resolution pixel values to generate new pixel values for the high-resolution image. A new technique based on an adaptive filter model is also being investigated for the super-resolution restoration of a continuous image sequences.

A local statistics based contrast enhancement technique for enhancing the reconstructed high resolution image from a set of shifted and rotated low resolution images is being developed in the Vision Lab. Planar shifts and rotations in the low resolution images are determined by a phase correlation approach performed on the polar coordinate representations of their Fourier transforms. The pixels of the low resolution images are expressed in the coordinate frame of the reference image and the image values are interpolated on a regular high-resolution grid. The non-uniform interpolation technique which allows for the reconstruction of functions from samples taken at non-uniformly distributed locations has relatively low computational complexity. Since bi-cubic interpolation produces blurred edges due to its averaging effect, the edges of the reconstructed image are enhanced using a local statistics based approach. The center-surround ratio is adjusted using global statistics of the reconstructed image and used as an adaptive gamma correction to achieve the local contrast enhancement which increases the image sharpness.

VIPS – Vision Lab Project Demonstrations