Comparison On Data Driven Algorithms To Maintain Color Constancy In Industrial Camera For High Intensity Environment

Abdul Rahman, Abdul Ghani (2012) Comparison On Data Driven Algorithms To Maintain Color Constancy In Industrial Camera For High Intensity Environment. Project Report. UTeM, Melaka, Malaysia. (Submitted)

	PDF (24 Pages) Comparison_On_Data_Driven_Algorithms_To_Maintain_Color_Constancy_In_Industrial_Camera_For_High_Intensity_Environment_-_24_Pages.pdf - Submitted Version Download (316kB)
	PDF (Full Text) Comparison_On_Data_Driven_Algorithms_To_Maintain_Color_Constancy_In_Industrial_Camera_For_High_Intensity_Environment.pdf - Submitted Version Restricted to Registered users only Download (2MB)

Abstract

Today, industrial camera had been widely used especially for monitoring and safety. But, under the high intensity environment such as strong daylight or the intense light of surgical rooms, industrial camera has difficulty maintaining the original essence of the image. In high intensity environment, the image will have a strong shadow, highlight, and the color of the scene will tend to deteriorate and influenced by the color of the high intensity illuminant. Therefore the need to maintain the original color of the scene is inevitable. In this project, two data-driven color constancy algorithms which are gray world method and retinex method are tested against the high intensity environment. The performance of these methods will be compared by analyzing the high intensity image with the image under a normal intensity environment by using a histogram and three types of error which is angular error, Euclidean error and root mean square error. The result shows that 75% of the angular error for the retinex method is lower than the angular error for the gray world method. Hence, retinex method will give a better result compare to the gray world method. Retinex method also can decrease the effect of highlight and remove the effect of strong shadow better than gray world method

Actions (login required)

View Item