A Detecting Algorithm Of Impurity In Medicinal Liquid Based On Video Sequences

The medical liquid has been widely used in the clinical treatment and most medical liquids get into human body by injecting intravenously. Because of some factors such as the medical liquid itself, production technology and production process, medical liquid may contain some unsolvable impurities, which will cause a lot of damage for the patient. Recently, the method of manual lamp inspection is widely used in our country. The result of this method is affected by lamp inspector seriously, too subjectivity. To develop an automatic detecting system of impurity in medicinal liquid is imminent, and it has the important theoretical significance and practical application value.The subject researches on the video sequence of medicinal liquid, and especially focuses on the impurity detecting algorithm of medicinal liquid automatic detecting system, which using high-speed video camera to get the test video sequence image, and appropriate image processing algorithm to extract and distinguish the impurity which is in the medicinal liquid, finally whether the medicinal liquid is eligible is automatically inferred.According to the function requirement and evaluating indicator of the detection, the whole design scheme of the detecting system is established. After introduced the system detecting principle, system composition and the function of each subsystem, a light source and lighting system is designed to meet the practical requirement and finally the appropriate camera and lens are chosen to get the image.As the large background noise exists in the original image, morphological filtering and Second-difference and energy accumulation methods are used to suppress the noise and enhance the contrast of image. Additionally, combined fuzzy c-means clustering algorithm based on spatial constrains which is used for segmenting image, an automatic detecting algorithm is put forwarded. The algorithm makes use of clustering criterion function which is established by image gray-scale information and spatial information to inhibit the noise influence of clustering center, finally the impurity is accurate extraction.There is similarity between bubble images which appeared when medical liquid circumrotate in high speed and impurity images .In order to distinguish bubbles and impurities, an identification algorithm is also put forwarded in the article, which based on the information of track and geometry. The algorithm adopts track association principles to get the hypothetical motion trail at first, then judge whether the impurity is exist via the continuity of the impurity moving trails and the randomness of the bubble noise moving trails, finally, whether the medical liquid is eligible is judged.The practicability of the proposed algorithm is proved by checking the video sequence image of gathered medical liquid with MATLAB7.0. The experiment results show that the algorithm which proposed in the paper can meet the requirements of detecting of impurity in medicinal liquid. The proposed algorithm can almost achieve the goal as expected.