Research Of Fabric Defects On-line Detection Method Based On Multi-cameras Machine Vision

In the clothing production, the quality detection of the fabric surface is an important prerequisite for the quality of products. At present, this work is mainly completed manually, which is low detection efficiency and bad reliability. So the automatic on-line detection system is needed. In this thesis, a fabric defects on-line detection system based on multi-cameras machine vision has been designed to meet this demand. Some related technologies have been studied to perform defects detection and obtain information of defects, and have been completed a defect access to information. This thesis has studied in theory and experiment, and the main research work are as the following:(1)This thesis has proposed a design scheme of detection system based on machine vision to perform every process in detection, comparing with manual detection.(2)This thesis has researched the algorithms of fabric defects detection. A detecting method based on Gabor filter, which has many calculations and low detection speed, has been improved, and a two grade method has been proposed to overcome these shortcomings; In addition, a method based on analysis of image energy has been proposed. The principle of this method is enhancing the contrast of texture and defect by filtering energy in main orientation, and then the defects will be detected.(3) This thesis has completed the calibration of detection system based on multi-cameras and defects location. It is first to complete calibration of single camera using a calibration plate with numbers. Then, figure out the relationship of the world coordinates between adjacent cameras, using a point in public region covered by two cameras. By now, data splice has been completed, namely the calibration of the whole system has been completed.(4)This thesis has completed the hardware design of system. Besides selection and design of the system's main hardware equipment, the mechanical device has been designed on the basis of a cloth inspecting machine. This device ensures that the system can obtain clear fabric images stably when it is in a certain speed transmission. At the same time, the yard measurer in cloth inspecting machine can obtain the distance signal of the fabric's translation, thus trigger the camera to image acquisition.(5)This thesis has completed the software design of system and has realized every functional module—image acquisition, defects detection, defects localization, and the storage of defect's information.