Enhancing micro-organism images by estimating image background
One- and two-dimensional techniques are proposed for enhancing micro-organism images using linear least-squares fitting, median filtering combined with bicubic approximation and wavelet filtering, respectively. The quality of enhanced images with pixel dimensions of 1200 x 1600 is examined from their corresponding binary images. It shows that one-dimensional modeling requires much less computation but provides relatively low quality output. On the other hand, the two-dimensional approximations for non-uniform lighting background, using both bicubic spline and wavelet filtering, are more effective and yield very high quality binary images tested with three different grayscale images.