아동 미술심리검사를 위한 AI 기반 그림 데이터 분석 모델 연구

국문 초록 (Abstract)

아동의 미술심리검사를 위하여 여러 가지의 분석 기법이 있으며, 그중 비교적 접근방법이 쉽고 AI 분석에 사용하기 쉬운 분석 기법으로 HTP (HouseTree-Person) 검사에 대하여 AI를 통하여 그림 심리상담사의 상담을 돕는 역활을 수행하기 위한 기초 연구이다. AI 모델은 아동이 그린 집, 나무, 사람 그림 중 집 그림에 대한 분석을 통해 아동 심리 상태를 추론한다. 현재 아동의 심리상담을 위하여 HTP를 수행할 수 있으나 이를 위해 특정 시간 에 해당 전문가와의 면담이 필요하다. 이때 여러 가지의 환경적 변수가 상담을 불편하게 할 수 있으며, 실제로 심리상담사의 전문적인 상담이 필요할 정도의 학 생은 2013년~2016년을 기준으로 초・중・고등학교 전체 학생의 약 3.9%의 비율 이다. 이러한 관심군 학생들을 선별하기 쉽도록 심리상담사의 상담을 돕는 것을 목표로 한다. 본 연구를 통하여 집에 대한 분석을 수행하고 향후 나무와 사람에 대한 분석 모델의 추가가 필요하다.

다국어 초록 (Multilingual Abstract)

There are various analysis techniques for children's art psychological tests. Among them, the analysis technique that is relatively easy to access and easy to use for AI analysis, which provides advice for the art psychological counselor is the HTP (House-Tree-Person) test. This is a basic research to perform a helping role. The AI model infers the child's psychological state through analysis of the house drawings among the drawings of houses, trees, and people. Currently, HTP can be performed for the children's psychological counseling but this requires an interview with the relevant expert at a specific time. By these environmental variables they can make the counseling uncomfortable, and in fact, the number of students who need professional counseling from a psychological counselor is approximately 3.9% of all elementary, middle, and high school students over the three years period from 2013~2016. The goal is to provide advice for the psychological counselor to make it easier to select students of this interest group. Through this study, it is necessary to conduct analysis on houses and will add analysis models for trees and people in the future.

목차 (Table of Contents)

• 국문초록 ···················································································· i
• 목 차 ··················································································· ii
• 표 목 차 ··················································································· v
• 그림목차 ··················································································· v
• 제 1 장 서 론 ·········································································· 1
• 제 1 절 연구의 배경 및 목적 ··········································· 1
• 1. 연구 배경 ····································································· 1
• 2. 선행연구의 한계점 및 연구목적 ····························· 1
• 제 2 절 연구 방법 및 논문 구성 ····································· 2
• 제 2 장 이론적 배경 ····························································· 3
• 제 1 절 미술 치료 ································································· 3
• 1. 미술 치료의 개념 ······················································· 3
• 2. 미술 치료의 장점 ······················································· 4
• 3. 미술 치료실의 환경 ··················································· 5
• 4. 치료 시간의 구성 ······················································· 5
• 제 2 절 미술 치료의 상징 ················································· 6
• 1. 미술 치료 상징의 이해 ············································· 6
• 2. 공간의 상징 ································································· 6
• 3. 선의 상징 ····································································· 7
• 4. 형태의 상징 ································································· 7
• 제 3 절 HTP의 이해 ·························································· 8
• 1. HTP 그림 검사 ·························································· 8
• 2. 실시 방법 ····································································· 8
• 3. 그림의 해석 ································································· 9
• 제 3 장 연구 설계 ····························································· 10
• 제 1 절 연구 모형 ···························································· 10
• 제 2 절 연구 가설 ···························································· 11
• 제 3 절 데이터 분류 ························································ 11
• 1. AI HUB의 그림 데이터의 분석정보 ··················· 11
• 2. LLM 분류 ·································································· 12
• 3. 설문 진행 ··································································· 17
• 제 4 장 결과 분석 ······························································· 18
• 제 1 절 AI 분석 결과 ······················································ 18
• 1. LLM 및 YOLOv5 모델 기반의 분석 ·················· 18
• 제 2 절 설문조사 결과 종합 ·········································· 32
• 제 5 장 결론 ········································································· 35
• 제 1 절 연구 요약 ···························································· 35
• 제 2 절 연구의 한계점 및 향후 연구 방향 ··············· 36
• 1. 연구의 시사점 ··························································· 36
• 2. 연구의 한계점 및 향후 연구 방향 ······················· 36
• 참고문헌 ·················································································· 38
• Abstract ·················································································· 39
• 설문지 ······················································································ 40