VRL an eye

Local image enhancement and Dynamic Range Compression


Dmitry V. Fedorov, Baris Sumengen, B.S. Manjunath


The dynamic range of modern systems employed for biomedical imagery is usually higher than the dynamic range of standard screen display devices used. This discrepancy leads to the problem in the tone mapping from the acquired high dynamic range (HDR) into the lower dynamic range (LDR) of print or screen. In confocal microscopy problems in tone mapping arise when there are areas with different fluorescent responses and certain regions might not be visible or suffer from a severe loss of contrast as shown in Fig. 1. Another problem is uneven illumination which is very common in light microscopy. Our method combines global and local approaches by applying mapping on small portions of the image (tiles). Using our framework different tasks can be achieved, such as: dynamic range compression, uneven illumination correction and automatic vignetting (light fall off) correction.

Example 1

A single plane image from a laser scanning confocal microscope of a 3 day detached cat retina section stained with TOPRO, a nuclear dye.

Img. 2a) Original image with non-uniform TOPRO staining in the ONL layer Img. 2b) Enhanced image ready for cell counter and texture analysis Img. 2c) Enhancement map

Example 2

Dynamic range compression, space and time enhancement. The photo-bleaching effect responsible for gradual fluorescence decay was corrected for the entire stack (30 frames) globally and locally in 12 bit images.

Img. 2a) Original image stack with 30 frames Img. 2b) Spatio-temporally enhanced images Img. 2c) Blue - input images and red - enhanced


This project is supported by the NSF Information Technology Research grant #0331697.


  1. Dmitry Fedorov, Baris Sumengen and B. S. Manjunath,
    "Tile-based framework for local enhancement of bio imagery"
    Technical Report, VRL, ECE, UCSB, Dec. 2005.
    [abstract] [PDF] [BibTex]

    Abstract preview: "We present a simple framework for image enhancement using local image information. The main idea is to divide the image into small tiles and individually enhance each of these tiles. Enhanced tiles ar..." [more]