Vision Defect Identification System (VDIS) using Knowledge Base and Image Processing Framework

Size: px

Start display at page:

Download "Vision Defect Identification System (VDIS) using Knowledge Base and Image Processing Framework"

Magdalen Flynn
5 years ago
Views:

1 Vishal Dahiya* et al. / (IJRCCT) INTERNATIONAL JOURNAL OF RESEARCH IN COMPUTER AND COMMUNICATION TECHNOLOGY Vol No. 1, Issue No. 1 Vision Defect Identification System (VDIS) using Knowledge Base and Image Processing Framework Mrs. Vishal Dahiya Sr. Assistant Professor IITE, Ahmedabad Dr. Priti Srinivasan Sajja Associate Professor, SPU, VVNagar Abstract: In this research paper, a novel system has been designed which is using two discipline of Computer science area which are Artificial intelligence and image Processing. Fuzzy sets and logic has been chosen from AI and image classification, filtering, matching & rendering has been used from IP. The rendering of the image has been done as per the specification given by the patient and then backend side image classification and matching has been done. Fuzzy rules are defined using fuzzy logic & fuzzy set; then by using rule base and database the inferences have been given by the system that is VDIS. Human perception is taken into consideration in both the areas to come on the final output of VDIP; the final output forwarded to the human expert to make final decision. Thermodynamics of mind [4,5,6,7]. Figure 1 is demonstrating the human perception process. There are number of factors which are affecting our perception which can be internal (Sensory Limits and Thresholds, Psychological Factors) and external factor (Target-stimuli and situation). Keyword : perception, rendering, rule, knowledge. 1. Introduction Human vision system starts with just the shower of photons that hit the retina of each eye and proceeds to construct a 2D contours and 3D shape by consulting various sources of information such as shading, texture, motion, occlusion, binocular disparity. In this process it uses many law which is based on reflectance, geometry, projection and lighting. Every aspect of a visual experience has its physiological counterpart in the nervous system called perception. The nature of these brain processes is such that they can be thought of as field processes where interaction between the parts and the whole are a general phenomenon. There are number of factors that effects human perception like visual weight (that is the combination of location, depth, size, color, shape, knowledge), Visual direction (that is weight direction, structural skeleton, movement), Top versus Bottom (that is environmental orientation, retinal orientation), Right versus left, Pseudo- Figure 1: Human Perception Process In this research paper, based on the human perception two approaches are used to find out the defects in human eye. The two approached uses different technique to discover the Vision defects, the eye of the patient may have. In the section 2, Knowledge Based VDIS has been discussed which is based on the Fuzzy set and Fuzzy theory. Section 3 would be discussing about the Image processing based VDIS that is using image classification, filtering, matching and lastly rendering the perceived image on the screen to provide the defects the patient eye may have. Section 4 will be showing the results of both approaches. Section 5 will be discussion about VDIS; which is followed by conclusion in Section Knowledge Based VDIS 5

A paradigm is a set of rules and regulations which defines boundaries and tells us what to do to be successful in solving problems within these boundaries.

2 Fuzzy Set Theory was formalized by Professor Lofti Zadeh at the University of California in 1965 [3]. What Zadeh proposed is very much a paradigm shift that first gained acceptance in the Far East and its successful application has ensured its adoption around the world. A paradigm is a set of rules and regulations which defines boundaries and tells us what to do to be successful in solving problems within these boundaries. For example the use of transistors instead of vacuum tubes is a paradigm shift - likewise the development of Fuzzy Set Theory from conventional bivalent set theory is a paradigm shift. In this section the knowledge base of the VDIS has been discussed. The KB of VDIS is consists of Fuzzy rule base and Database. The fuzzy rule base consists of three layer of fuzzy rule set and three layer of Fuzzy inferences described in Figure 2. Each Rule in the rule base have three input and all the combination of the three input has been implemented which gives 343 combinations as an output; these 343 combination will be acting as input to inference engine and one output. This one output is an input to the Fuzzy rule set Layer 2. The same process continues till one final output is provided by the system. Each Rule in the rule base have three input and all the combination of the three input has been implemented which gives 343 combinations as an output; these 343 combination will be acting as input to inference engine and one output. This one output is an input to the Fuzzy rule set Layer 2. The same process continues till one final output is provided by the system. define the problem and due to uncertainty in the situations in which that particular problem occurs. Probability theory is an age old theory, which excellently handles this uncertainty. But this probability theory can be applied only to situations whose occurrence of events is greatly determined by random process [3, 8]. Fuzzy logic is a multi valued logic address the word approximate rather than to be precise. In real world situation we often face problem with vague information. This vague information s are easily understandable by human beings but are hard to interpret computationally. For example weather forecasting, decision making with uncertain conditions, diagnosing problem, speech recognition, image processing etc.[1,2,9] Figure 3 is showing one Fuzzy rule set that is rule 1 and also the inference which will be output from the fuzzy inferences layer 1. Figure 3: Rule 1 of VDIS Figure 2: Fuzzy rule and inferences in Vision defect Identification System(VDIS) 2.1 Database for KB There are many tables which have been designed for KB of VDIS. In this only two tables has been shown which are as Table 1 consists of all the questions and Table 2 consist of all the rules which are used for creating the inferences from the input values and coming on the final result. Table 1: List of Questions Problems in the real world turn out to be quite complex, due to uncertainty in the parameters that 6

Figure 4: Question List 1 Table 2: List of Rules Figure 5: Question List 2 Figure 6: Final Result 2.

Group of question has been considered and depending upon the uncertainty of the response of the patients-seven possible values are considered then fuzzy rule

Mapping of the response has been done with the database to come on the final decision. Figure 4, 5 is demonstrating the two groups of questions.

There are some specified effects which are applied on the image as per the details given by the patient; the operator applies those effects by selecting the

After applying all the changes, the image stores in the database of the patient with a name that is perceived image.

3 Figure 4: Question List 1 Table 2: List of Rules Figure 5: Question List 2 Figure 6: Final Result 2.2 User Interface The user interface is created in Java Server Pages. Group of question has been considered and depending upon the uncertainty of the response of the patients-seven possible values are considered then fuzzy rule base has been designed based on those question sets and the uncertain response of the user. Mapping of the response has been done with the database to come on the final decision. Figure 4, 5 is demonstrating the two groups of questions. Figure 6 will be showing the result that is most probable vision defects. 3. Image Processing Based VDIS An image will be the input to the system. There are some specified effects which are applied on the image as per the details given by the patient; the operator applies those effects by selecting the specified criterion based on the 0-1 scale. After applying all the changes, the image stores in the database of the patient with a name that is perceived image. Once the perceived image stored, then the image matching function is to be called to perform the comparison between perceived image and the disease images which are existing in Disease Image database. The nearest matching image is to be selected and the vision defect information of that selected image is to be mapped. Finally the vision defects name and detail displayed on the screen, which is helpful in identifying the problems a patient eye may have. This will be helpful to a doctor to get a decision. Figure 7 is showing the system structure. 7

3.2 User Interface of IP-VDIS Figure 7: System Structure of IP-VDIS 3.

Table 3 consists of Effect details. Table 4 is the Effect Priority table.

Figure 7(a-d) is showing the effects which is implemented on the original image and then right hand side image is

The perceived image is compared with the original image.

4 3.2 User Interface of IP-VDIS Figure 7: System Structure of IP-VDIS 3.1 Database There are many relations and tables which are designed but here only two of the tables has been shown; Table 3 consists of Effect details. Table 4 is the Effect Priority table. Table 3: Effect Details This application is developed using Java Swing & java 2D as front end and MySql as backend. Figure 7(a-d) is showing the effects which is implemented on the original image and then right hand side image is effected by the effects which is to be stored as perceived image. The perceived image is compared with the original image. The nearest matching in the database will be retrieved and the defect which is stored with that perceived image will be displayed on the output screen. Figure 8 is demonstrating the Defect Information Screen. Figure 7(a): Applied Effects Screen Table 4: Effect priority Figure 7(b): Applied Effects Screen 8

5 Figure 7(c): Applied Effects Screen Figure 7(d): Applied Effects Screen In this discussed VDIS, more than one disease can be suggested depending upon the symptoms observed by the doctor in the patient. Symptoms, matching will be done in the knowledge base and based on the reasoning defined in the knowledge base the vision defect information will be suggested to the doctor and doctor takes the final decision. This system is very useful to the doctors and the suggestion provided to the doctor which reduces the decision making timing taken by the doctor. This system is implemented and results of this have been shared in this paper. Fuzzy set theory and rule base plays a very important role in designing the human like intelligence and results are more efficient as compared to any other branch of Artificial Intelligence. As now a day s most of the systems are intelligent in nature and everyone prefer to use the system which has some or other decision making capability. So this system design is an addition to that list. 2 Conclusion Figure 8: Defect Information Screen 4 Discussions Both the approaches which are discussed in this paper give efficient results. VDIS is the merging of both the approaches. Same database has been shared by both approaches that is created using MySql. Human perception plays an important role in identification of the defect because that is the patient understanding that what symptoms, he will observed and tell to the operator, then operator applied those response in the Systems First the patient database has been created then the evaluation of the disease can be done. The registered patient can be diagnosis with the help of the designed system. The main interface of the system will be diagnosis the disease of the patient with the help of rule base reasoning. The discussed Vision Defect Identification System is using logical reasoning base entities for decision making which are approximately formulated with the help of fuzzy set theory and rule base. It requires more time to formulate the fuzzy set and rule base for any medical application. Image Processing concepts add more effectiveness to the final decision making because it s very easy by a patient to narrate his visions problem with respects to the effects which he may be facing in the original view and then the system also provide the range between 0-1 which gives more accuracy to the decision given by the system This system is very useful to finding the vision defect. It is implemented with the help of Java Technologies. This system is very helpful to the experts by giving an option for final decision making. Though many other diagnosis systems have been designed [10] and discussed but this is the novel diagnosis system which is the blending of two major fields of computer science area that is AI & IP. 3 References: [1] Priti Srinivas Sajja, and Jeegar A Trivedi, Using Type-2 Hyperbolic Tangent Activation Function in Artificial Neural Network, Research Lines, vol. 3, no. 2, pp , [2] WU, H. AND MENDAL, J.M., Uncertainty bounds and their use in the design of interval type-2 fuzzy logic system. IEEE Transactions on fuzzy systems, vol. 10, no. 5, pp ,

6 [3] Zadeh, L.A Fuzzy sets, Information and Control, Vol. 8, [4] L. T. Sharpe, A. Stockman, H. J agle, and J. Nathans. Color Vision: From Genes to Perception, chapter Opsin genes, cone photopigments, color vision, and color blindness, pages Cambridge University Press, [5] Rodrigo Palma-Amestoy, Edoardo Provenzi, Marcelo Bertalmı o, and Vicent Caselles A Perceptually Inspired Variational Framework for Color Enhancement IEEE Transactions on Pattern Analysis and Machine Intelligence, VOL. 31, NO. 3, MARCH [6] G. Wyszecki and W. S. Stiles. Color Science: concepts and methods, quantitative data and formulae. John Wiley and Sons, 2 nd edition, [7] S. J. Jerome Teng, Robust Algorithm for Computational Color Constancy IEEE conference proceeding on title International Conference on Technologies and Applications of Artificial Intelligence, pp 1-8, Dec [8] Chang, C.-W.; Ying, H.; Hillman, G.R.; Kent, T.A. and Yen, J. (1998). A rule -based fuzzy segmentation system with automatic generation of membership functions for pathological brain MR images. Computers and Biomedical Research, html [9] Priti Srinivas Sajja Fuzzy artificial neural network decision support system for course selection, Journal of Engineering and Technology, vol.19, pp [10] Chin-Lun Lai, Shu-Wen Chang An Image Processing Based Visual Compensation System for Vision Defects Pages , IEEE conference processing Nov

Perception. What We Will Cover in This Section. Perception. How we interpret the information our senses receive. Overview Perception

Perception. What We Will Cover in This Section. Perception. How we interpret the information our senses receive. Overview Perception Perception 10/3/2002 Perception.ppt 1 What We Will Cover in This Section Overview Perception Visual perception. Organizing principles. 10/3/2002 Perception.ppt 2 Perception How we interpret the information