A Wavelet Based Complementary Approach for Image Enhancement

Detail in an image means more meaningful information that is very important in many computer vision and pattern recognition applications. The object region visibility in an image plays an important role in obtaining accurate and desired information from the original image. In particular, image processing techniques developed for region segmentation and object classification have better results depending on the visibility in images. There are several enhancement techniques available which are capable of obtaining clear images with balanced lighting and contrast. In this thesis, a completely image dependent approach to enhance the luminance of images under extreme lighting conditions and a wavelet based adaptive contrast enhancement algorithm to enhance faint image features is proposed. Additionally, a wavelet based denoising algorithm is used to eliminate the noise in the processed images. A specifically designed nonlinear transfer function is used to enhance the luminance of the image. All the parameters required for controlling the curvature of the nonlinear function are computed based on the image statistics. The wavelet coefficients are modified by a transfer function whose parameters are also dependent on the image statistics in the contrast enhancement algorithm. Experiments performed on several images show that the proposed techniques are capable of providing better visibility in images along with enhanced high frequency details.