영상처리기법을 이용한 분무액적 크기의 측정

Spray droplet size measurement using image processing technique

본 연구에서는 oil-bath를 이용한 종래의 방법에 의해 얻어진 사진으로 부터 분무액적의 수와 크기를 자동적으로 측정하는 방법을 시도하였다.특히 측정을 효과 적으로 수행하기 위해 한 영상에서 비교적 맣은 분무액적을 측정하는 방법과 붙어있는 액적들이나 이물질 등을 걸러내는데 중점을 두었으며, 아울러 본 연구를 통하여 설정 된 측정방법의 정확도를 검토하였다.

An economic and efficient system for measuring drop sizes was developed. Pattern recognition technique was used with conventional oil-bath sampling and photographic method. The system was designed to measure and count relatively large number of drops in a very short time, and also to filter out abnormal images such as drops in contact or overlap as well as odd-shaped foreign materials. In this measuring system, most important error originates from the process of converting the original image to the binary image. If the photograph contains a large number of spray drops, the relative size of the pixel to the drops is not infinitesimally small; thus the proper choice of threshold level to convert the original image to the binary image becomes very important. In present case, most of the images lay in one of the two separate bands of brightness level and the arithmetic mean of the most popular brightness levels from each band was chosen as the threshold level. Present image processing system reduces the subjective error by the observers in counting and measuring drops and turns out to be substantially effective. The processing time can be further reduced by improving the hardware system concerned with the digital image coding.