The Crystal Counter Software Is Designed

In the manufacturing process of LED, one of the working procedures is to count the tube chip. General counting method is: getting the enlarged digital image of tube chips in a certain proportion by photographing or scanning them, and then counting with the help of digital image processing technology. The digital image procedures include three parts: image pre-processing, image partitioning and profile description. Image partitioning, as the most important part, has a direct impact on the result of the following profile description work. Pattern recognition method is often needed to deal with the complicated image partitioning work. If the image is composed of several abstract objects that are located in a high contrast background, we can use grey-value division algorithm to partition the preprocessed image directly. In this way, grey-value division algorithm is a key factor for counting precision. The significance of this research lies in the calculating of tiny particles with computer system. There are lots of similar applications, such as counting cells. Algorithm is studied in this thesis to give a guide for the following applications. In image processing systems, it is very easy to obtain the image of an object with instruments. We can also get different images, such as infrared image and X-ray image, in a low cost and in variable environments. The exact quantity of tested objects is often needed in many applications. The research in this thesis lays a solid theoretical foundation for those image procession related applications. Here we mainly talk about how to deal with a image on which there are lots of crystal points and fuzzy borders. Before the image is transmitted into 2-value-image, it needs lots of operation such pre-process, erase noises. In our common work, a image always has lots of undetermined factor, so we will have to do a lot of work on pre-process according to the character of the image .Here we show the common way of pre-processing and some special pre-process way about our image. We also talk about some matrixes, every matrix has its own filter functions. Some are used for erasing noising and some for strengthening image .We have to erase the noises first, then we can get the valve value. How to get the true valve value is discussed more than anything others, because the valve value has directly relationship with the 2-value image and the counting algorithm. If we get the 2-value image, there are many kinds of algorithm to realize the function of counting the crystal points. But most of them are too slow to be accepted by customer. This traditional algorithm trend to scan the whole pictures one point by one point, which occupy lots of time of CPU. The algorithm we used here can get the result of the crystal points within a short time. The key point we are talking about is the pre-processing to an image and determination of the valve value. You will find that the algorithm is relatively easy, but it is very hard to determine the valve of a picture. The counting algorithm is the key to the target, which is very important to our customers. We spent lots of time and energy on it during debugging our software. The results show that if we handle a picture of BMP format which is about 1.76MB and has more than 4000 crystal points, we can get the precise result with error rate under 2‰within 5 seconds .To tell you the truth, we find the error rate by manual work .According to the result we can draw the conclusion that our customer will be satisfied with our system. More than 80% information that people acquire from our world come from seeing ,so image processing will be widely used in the future .In counting system field, people can enjoy not only higher efficiency but higher precision .At the same time, in research field, image processing is destined to be used widely. By our research, we will deal with this kind of question more easily and handle relative question more quickly. Finally, we will realize the fact that technology is transmitted into productivity.