The Surface Defects Of Lithium Electrode Detection System Based On DSP

As we know that the quality of the lithium electrode plays a decisive role inlithium battery’s useful life, so it is necessary to focus on a "real-time defectdetection system" in the production process of lithium electrode. A "real-timedefect detection system" in the production process of lithium electrode would makea fast and accurate identification for the surface defect of lithium electrodesso the result can be commendably classified.That could save manpower and provideguarantee to improve the quality of the lithium electrode.Then we would make up the surface defects of lithium electrode hardwaredetection system based on DSP according to the requirements of lithium electrodeproduction. And we chose a fit way of the strip-shaped LED light source and a frontand rear diffuse illumination method for the lithium electrode in view of theanalysis of the characteristics of the light source.We use a TCD1501D line arrayCCD camera and a camera lens of75mm focal length,and make a design for theinterface of circuit between camera and digital signal processor.Then we choseTMS320DM648as the core of the image processing,and FLASH and SDRAM as programand data memory chip.Through the study of the filtering algorithm,by using an improved medianfiltering algorithm,we can reduce the running time of the algorithm.We also analysethe the causes of the errors of Blob Analysis in the surface defects detection oflithium battery electrodes.So we think that a traditional method of thresholdsegmentation is not fit for the surface defect detection of lithium batteryelectrodes.In this paper,we make an improved Canny algorithm based on edgesegmentation instead of threshold segmentation.As the collision between thesmoothing scale and detection precision of the traditional Canny algorith,we usean adapative smoothing parameters.we use adaptive threshold selection methodwhich based on divided images instead of artificial selection of traditional gradientthreshold. So wiht a improved Canny algorithm and Blob Analysis we can solve theproblems in the defect detection of lithium battery electrode surface which caused by the threshold segmentation.And we have the analyses and choices in the features of the surface defectsof lithium battery electrode to make a decision tree classification method andsystem for the defects.Those detects are divided into four types as scratch, foilexposed, cracks and dirt.At last we would achieve and transplant the algorithm ona experimental platform based on DSP.When the results comes out it will bedisplayed and storaged in real-time,that’s the final goal.