Transform domain measure of enhancement — TDME — For security imaging applications

Today, many security applications rely on imaging sensors. However, the quality of the captured image is highly susceptible to environmental lighting conditions such as poor or non-uniform illumination. Security imaging systems rely on efficient real-time image enhancement. For autonomous systems, determining and producing the best visually enhanced image output as perceived by the human visual system remains a challenge. In this paper, we present a metric for autonomous evaluation that will enable security systems to automatically determine the best human visual quality image. To achieve this, we have established a “no parameter no reference” metric that can determine the best visually pleasing image. The metric is algorithm independent such that it can be utilized for a diversity of enhancement algorithms. We present our DCT transform domain measure of enhancement (TDME). Unlike spatial domain measure of enhancement methods, the proposed measure is independent of image attributes and does not require any parameter selections to operate. The proposed measure is applicable to compressed and non-compressed images and can be used as an enhancement metric in conjunction with many different image enhancement methods.