An Efficient Technique for Non-Uniformity Correction of Infrared Video Sequences with Histogram Matching

• Published in: Journal of electrical engineering & technology 2022, 17(5), , pp.2971-2983
• Main Authors: Abbass, Mohammed Y., Sadic, Nevein, Ashiba, Huda I., Hassan, Emad S., El-Dolil, Sami, Soliman, Naglaa F., Algarni, Abeer D., Alabdulkreem, Eatedal A., Algarni, Fatimah, El-Banby, Ghada M., Abdel-Rahman, Mohamed R., Aldosari, Saeed A., Dessouky, Moawad I., El-Rabaie, El-Sayed M., El-Shafai, Walid, Khalaf, Ashraf A. M., El-Dokany, Ibrahim M., El-Samie, Fathi E. Abd
• Format: Article
• Language: English
• Subjects: Electrical Engineering; Electrical Machines and Networks; Electronics and Microelectronics; Engineering; Instrumentation; Original Article; Power Electronics; 전기공학
• ISSN: 1975-0102; 2093-7423
• DOI: 10.1007/s42835-022-01010-9

Infrared (IR) image sequences are acquired with certain types of cameras. These cameras give the sequence of images according to the heat distribution. With time, some deterioration of the quality of the sequence occurs due the thermal noise effect generated in the camera. This thermal noise effect leads to some sort of non-uniformity in the obtained image sequence. Hence, it is necessary to perform some sort of non-uniformity correction in the video sequence according to the first frame. This type of non-uniformity correction is scene-based. This paper introduces a scene-based non-uniformity correction technique that depends mainly on histogram matching. The noise effect on each frame in the sequence leads to some drift in the histogram of that frame. Hence, the proposed technique depends on the histogram matching concept to correct the histogram of each frame in the sequence based on the histogram of the first frame that is free from the thermal noise effect. Different image quality metrics including entropy, contrast, edge intensity, average gradient, and correlation with the first frame are adopted to assess the quality of the obtained frames after adjustment. It is required in the frames to be corrected to reduce entropy, edge intensity and average gradient as these metrics are increased with the presence of thermal noise effect on all pixels represented as much details and unnecessary information. In addition, the contrast of the video sequences should be increased to determine objects in a better way. The correlation of the corrected frames with the first one should be increased to reduce the noise effect. Simulation results reveal enhanced quality of the obtained video sequences after processing with the proposed technique.