• 제목/요약/키워드: gray level

검색결과 702건 처리시간 0.021초

형태학과 문자의 모양을 이용한 뉴스 비디오에서의 자동 문자 추출 (Automatic Text Extraction from News Video using Morphology and Text Shape)

  • 장인영;고병철;김길천;변혜란
    • 한국정보과학회논문지:컴퓨팅의 실제 및 레터
    • /
    • 제8권4호
    • /
    • pp.479-488
    • /
    • 2002
  • 최근 들어 인터넷 사용의 증가와 더불어 디지털 비디오의 수요 또한 급격히 증가하고 있는 추세이다. 따라서 디지털 비디오 데이타베이스의 인덱싱을 위한 자동화된 도구가 필요하게 되었다. 디지털비디오 영상에 인위적으로 삽입되어진 문자와 배경에 자연적으로 포함되어진 배경문자 등의 문자 정보는 이러한 비디오 인덱싱을 위한 중요한 단서가 되어질 수 있다. 본 논문에서는 뉴스 비디오의 정지 영상에서 뉴스 자막과 배경 문자를 추출하기 위한 새로운 방법을 제안한다. 제안된 알고리즘은 다음과 같이 세 단계로 구성된다. 첫 번째 전처리 단계에서는 입력된 컬러 영상을 명도 영상으로 변환하고, 히스토그램 스트레칭을 적용하여 영상의 수준을 향상시킨다. 이 영상에 적응적 임계값 추출에 의한 분할 방법을 수정 적용하여 영상을 분할한다. 두 번째 단계에서는 적응적 이진화가 적용된 결과 영상에 모폴로지 연산을 적절하게 사용하여, 우선 문자 영역은 아니면서 문자로 판단되기 쉬운 양의 오류(false-positive) 요소들이 강조되어 남아있는 영상을 만든다. 또한, 변형된 이진화 결과 영상에 모폴로지 연산과 본 논문에서 제안한 기하학적 보정(Geo-corrertion) 필터링 방법을 적용하여 문자와 문자로 판단되기 쉬운 요소들이 모두 강조되어 남아있는 영상을 만든다. 이 두 영상의 차를 구함으로서 찾고자 하는 문자 요소들이 주로 남고, 문자가 아닌 문자처럼 보이는 오류 요소들은 대부분 제거된 결과 영상을 만든다. 문자로 판단되는 양의 오류 영역들을 남기는데 사용된 모폴로지 연산은 3$\times$3 크기의 구조 요소를 갖는 열림과 (열림닫힘+닫힘열림)/2 이며, 문자 및 문자와 유사한 요소들을 남기는데 사용된 연산은 (열림닫힘+닫힘열림)/2와 기하학적 보정이다. 세 번째 검증 단계에서는 전체 영상 화소수 대비 각 후보 문자 영역의 화소수 비율, 각 후보 문자 영역의 전체 화소수 대비 외곽선의 화소수 비율, 각 외곽 사각형의 폭 대 높이간의 비율 등을 고려하여 비문자로 판단되는 요소들을 제거한다. 임의의 300개의 국내 뉴스 영상을 대상으로 실험한 결과 93.6%의 문자 추출률을 얻을 수 있었다. 또한, 본 논문에서 제안한 방법으로 국외 뉴스, 영화 비디오 등의 영상에서도 좋은 추출을 보임을 확인할 수 있었다.

구충증(鉤蟲症)에 관(關)한 연구(硏究) 제1편(第1篇) 구충(鉤蟲)의 감염(感染) 및 구충성빈혈(鉤蟲性貧血)에 관(關)한 고찰(考奈) (Studies on Ancylostomiasis I. An Experimental Study on Hookworm Infection and Anemia)

  • 이문호;김동집;이장규;서병설;이순형
    • 대한핵의학회지
    • /
    • 제1권1호
    • /
    • pp.55-66
    • /
    • 1967
  • In view of its prevalence in the Far East area, a more detailed knowledge on the hookworm infection is one of the very important medical problems. The present study was aimed to; determine the infectivity of the artificially hatched ancylostoma duodenale larvae in man after its oral administration, evaluate the clinical symptomatology of such infection, determine the date of first appearance of the ova in the stool, calculate the blood loss per worm per day, assess the relation-ships between the ova count, infectivity(worm load), blood loss and severity of anemia. An erythrokinetic study was also done to analyse the characteristics of hookworm anemia by means of $^{59}Fe\;and\;^{51}Cr$. Materials and Methods Ten healthy male volunteers(doctors, medical students and laboratory technicians) with the ages ranging from 21 to 40 years were selected as the experimental materials. They had no history of hookworm infection for preceding several years, and care was taken not to be exposed to reinfection. A baseline study including a through physical examinations and laboratory investigations such as complete blood counts, stool examination and estimation of the serum iron levels was done, and a vermifuge, bephenium hydroxynaphoate, was given 10 days prior to the main experiment. The ancylostoma duodenale filariform larvae were obtained in the following manner; The pure ancylostoma duodenale ova were obtained from the hookworm anemia patients and a modified filter paper method was adopted to harvest larger number of infective larvae, which were washed several times with saline. The actively moving mature larvae were put into the gelatine capsules, 150 in each, and were given to the volunteers in the fasting state with 300ml. of water. The volunteers were previously treated with intramuscular injection of 15mg. of chlorpromazine in order to prevent the eventual nausea and vomiting after the larvae intake. The clinical symptoms and signs mainly of the respiratory and gastrointestinal tracts, appearance of the ova and occult blood in the stool etc. were checked every day for the first 20 days and then twice weekly until the end of the experiment, which usually lasted for about 3 months. Roentgenological survey of the lungs was also done. The hematological changes such as the red blood cell, white blood cell and eosinophil cell counts, hemoglobin content and serum iron levels were studied. The appearance of the ova in the stool was examined by the formalin ether method and the ova were counted in triplicate on two successive days using the Stoll's dilution method. The ferrokinetic data were calculated by the modified Huff's method and the apparent half survival time of the red blood cells by the modified Gray's method. The isotopes were simultaneously tagged and injected intravenously, and then the stool and blood samples were collected as was described by Roche et al., namely, three separate 4-day stool samples with the blood sample drawing before each 4-day stool collection. The radio-activities of the stools ashfied and the blood were separately measured by the pulse-height analyser. The daily blood loss was calculated with the following formula; daily blood loss in $ml.=\frac{cpm/g\;stool{\times}weight\;in\;g\;of\;4-day\;stool}{cpm/ml\;blood{\times}4}$ The average of these three 4-day periods was given as the daily blood loss in each patient. The blood loss per day per worm was calculated by simply dividing the daily blood loss by the number of the hookworm recovered after the vermifuge given twice a week at the termination of the experiment. The iron loss in mg. through the gastrointestinal tract was estimated with the daily iron loss in $mg=\frac{g\;Hgb/100ml{\times}ml\;daily\;blood\;loss{\times}3.40}{100}$ 3.40=mg of iron per g Hgb following formula; Results 1. The respiratory symptoms such as cough and sputum were noted in almost all cases within a week after the infection, which lasted about 2 weeks. The roentgenological findings of the chest were essentially normal. A moderate degree of febril reaction appeared within 2 weeks with a duration of 3 or 4 days. 2. The gastrointestinal symptoms such as nausea, epigastric fullness, abdominal pain and loose bowel appeared in all cases immediately after the larvae intake. 3. The reduction of the red blood cell count was not remarkable, however, the hemoglobin content and especially the serum iron level showed the steady decreases until the end of the experiment. 4. The white blood cells and eosinophil cells, on the contrary, showed increases in parallel and reached peaks in 20 to 30 days after the infection. A small secondary rise was noted in 2 months. 5. The ova first appeared in the stool in 40. 1 days after the infection, ranging from 29 to 51 days, during which the occult blood reaction of the stool became also positive in almost cases. 6. The number of ova recovered per day was 164, 320 on the average, ranging from 89,500 to 253,800. The number of the worm evacuated by vermifuge was in rough correlation with the number of ova recovered. 7. The infectivity of ancylostoma duodenale was 14% on the average, ranging from 7.3 to 20.0%, which is relatively lower than those reported by other workers. 8. The mean fecal blood loss was 5.78ml. per day, with a range of from 2.6 to 11.7ml., and the mean blood loss per worm per day was 0.30ml., with a range of from 0.13 to 0.73ml., which is in rough coincidence with those reported by other authors. There appeared to exist, however, no correlation between the blood loss and the number of ova recovered. 9. The mean fecal iron loss was 2.02mg. per day, with a range of from 1.20 to 3.89mg., which is less than those appeared in the literature. 10. The mean plasma iron disappearance rate was 0.80hr., with a range of from 0.62 to 0.95hr., namely, a slight accerelation. 11. The hookworm anemia appeared to be iron deficiency in origin caused by continuous intestinal blood loss.

  • PDF