Research On Fabric Color Difference Detection Algorithm Based On Color Appearance Model

With the continuous improvement of consumer requirements for fabric color quality,fabric color difference detection has become one of the important issues faced by textile enterprises.Traditional fabric color difference detection mainly relies on manual observation and color difference instrument detection.However,manual observation methods are highly subjective and result in inaccurate results,while color difference instrument detection requires high levels of worker operation and slow detection speed.In order to solve these problems,machine vision technology is widely used in fabric color difference detection.By collecting fabric images using an industrial camera and combining detection algorithms,color difference detection of the fabric is achieved.At present,most fabric color difference detection algorithms are based on camera captured images for color difference detection,ignoring the impact of environmental factors on the color of fabric images.This leads to inaccurate algorithms in practical applications that are easily affected by environmental factors.Therefore,this article proposes an algorithm that considers different environmental factors and can effectively achieve fabric color difference detection.Firstly,representative fabrics were selected as the research object for analysis,and suitable industrial cameras and lenses were selected based on their characteristics.At the same time,a fabric image acquisition platform was established.Considering the requirement of fabric width,a multi camera image acquisition mode has been designed to ensure the integrity and accuracy of the image.Secondly,a fabric image preprocessing algorithm was studied for the collected fabric images.The Gaussian filter algorithm is used to remove the noise generated in the process of cloth image acquisition,and the automatic threshold method is used to remove the redundant edges of the image to extract the complete cloth image.For the images captured in the multi camera mode,the scale invariant feature transformation is used for image registration,and the weighted average method is used to eliminate the image seam to obtain the cloth image with full width.Then,a color appearance model applied to fabric color difference detection was studied.By calculating key parameters,a CAM16 color appearance model and a CAM16-USCS uniform color space were established,achieving the calculation of fabric color appearance attributes.Subsequently,the color appearance attributes were input into a uniform color space and the fabric color difference was calculated using the color difference formula.At the same time,the results of the color appearance model,CIELAB,CMC(l:c),and CIEDE2000 color difference formula were compared,proving the rationality of the color appearance model applied to fabric color difference detection.Finally,the performance of the fabric color difference detection algorithm was tested.Human visual color difference experiments were conducted on fabric samples using the gray scale method to obtain visual color difference values.The performance of fabric color difference detection algorithm based on color appearance model was evaluated using performance factor method.At the same time,the standard residual sum of squares method was used to compare the advantages and disadvantages of CAM16 color appearance model,CIELAB,CMC(l:c),and CIEDE2000 color difference formula in fabric color difference detection.The results show that the performance of the color difference detection algorithm based on the color appearance model is superior to the other three methods.In order to facilitate the use of color difference detection algorithms,the software interface and overall architecture of the fabric color difference detection system have been designed.