Illumination correction of dyed fabrics method using the kernel extreme learning machine based on the improved Aquila Optimizer

Different environmental light affects the effect of textile surface color rendering, thus affecting the accuracy of machine vision-based textile quality inspection. In order to solve this problem, we propose an illumination correction model based on the improved Aquila Optimizer (AO) for the kernel extreme learning machine (KELM). First, a simulated annealing algorithm is used, and its better solution is used as the initial population of the AO, which increases the global search capability and improves the stability of the algorithm. Second, the KELM model is introduced and the AO optimization algorithm is used to obtain the optimal penalty parameter <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" display="inline" overflow="scroll"> <mml:mi>C</mml:mi> </mml:math> and kernel parameter γ of the KELM, thus constituting the SA-AO-KELM model. Finally, some textile images are selected to test the effect of the constructed light correction model using their chromaticity features, and the effect of light in the image is eliminated according to the model output. In order to verify the effectiveness of the designed illumination correction method, this paper adopts the chromatic aberration detection for the comparative analysis of the images before and after correction to evaluate the effect of illumination correction on improving the performance of textile color quality detection. The experimental results show that the method proposed in this paper can effectively improve the textile image presentation. The average color difference between the corrected image and the standard image is calculated to be 2.9991 using CIEDE2000, which reduces the color difference by about 24.43% compared with the traditional kernel extreme learning machine.