Research On Capsule Surface Inspection Based On Image Processing

With the development of the economy of our country, problems for the security of food and medicine arise more and more. Unqualified capsules are flowing into our market, which intimidate people’s healthy. So everyone demand our enterprise should provide medicine with a higher quality. high measures should be taken in order to guarantee the quality of products. while in china there is no equipments which can inspect the capsule to differ the qualified and unqualified capsules. In this circumstance, our team research the equipment that we called as Capsule Inspection Base On Machine Vision. In product line,worker can tell the qualified and unqualified capsules only by their eyes. There are some disadvantages for this. First of all, after worker use eyes in a long time, they will feel tired, which maybe make adverse impacts on their healthy eyes. So this way in which workers inspect the capsules by eyes not only lead a unreliable result but also arise leak to inspect unqualified capsules. the antiquated way never satisfy the strong demands from enterprises. In this paper,we will resolve this problem. the solutions I will describe are as followed. Building the image system is the first step. Our system consists of two industry cameras,two 1394 card and software which can sample images at the rate of 30/ps. the import part I want mentioned is how to project the light source. Doing a lots of experiments, finally we found that using the infrared to lighten up the capsule in the direction from the bottom to top layer. In the second step, we should design the algorithm to accomplish targeting objects, segment of asymmetry image, recognizing the defects in the image and so on. those actions above all get involved some high level image processing machine vision, such as edge detection, dynamic segment,pattern matching, fast line region growing. spot analysis and so on. This algorithm can quantize the defect in high precise by areas and the contrast with the background based on gray value. The third step I want to discuss in this paper is how to make synchronization between machine and the software which embed algorithms and image sample. We just use a single chip MCU 8051 to resolve this problem with high speed.The last important thing I want to tell you is software architecture which determine the efficiency of the algorithm. The architecture is so ingenious that can process four capsules images in simultaneity in which way the time algorithm take will decrease to half of part that before. as before time-consuming is 60ms, but now the time will not exceed 30ms. So time-consuming for complicate algorithm in image processing is resolved by parallel computing...