Designing And Fabricating Auto Lens Meter Base On Subpixel Algorithm

In the modern automatic industrial system, we have to deal with all kinds of detection, inspection and recognition. It is called the machine vision that the measured target is measured, identified, classifed, be position Using optical image. The design method of the automatic lens meter is use COMS image sensor to capture the target's image, then the image is be processed that it is filtered noise, edge detection and character be got. So we can get the lens parameter which is include sphere's degress, cylinders's degrss and prism's degree by the corresponding target image's dimension based on the polynomial interpolation method of sub-pixel segmentation algorithm.A novel algorithm to estimate locations and sizes of circles or ellipses to sub-pixel accuracy is proposed. The machine vision system combined with COMS LM9617, CPLD EPM3256, and ARM chip S3C44B0X as the image sensor, microprocessor, and the basic processing unit respectively. In addition, it also describes the colored liquid crystal LM057QC1T01 display and the design proposal of sensitive printer. In view of the request of the real time of the measurement system, we select NS Corporation's COMS image sensor LM9617, array structure of 640×480, and measurement scope of±2500°of ball mirror, therefore we must guarantee 5000 level distinguish precision through the precise sub-pixel algorithm. It is used that the polynomial interpolation method of sub-pixel segmentation algorithm to improve the measuring accuracy, Sobel operator with direction is used to change image edge gray scale into gradient images and fix the gradient direction of images in the boundary pixel, then the polynomial interpolation method of sub-pixel segmentation algorithm along the gradient direction is applied. The experimental result proved this algorithm can achieve 0.186 pixel precision.