Abstract: Now days, image searching technique is popularized on mobile phones where mobile users search and find desired images through their mobiles. But it is not easy to find desired image through mobile, because mobile phone’ screen is too small to display full image. It is also difficult for users to provide input on mobile phones to find targeted image. Hence, by considering all these issues, we have developed one image searching technique called – I2MMS: An Interactive Multi Modal Visual Search Technique. In text based technique, if user provides lengthy queries to find desired image then it will be difficult for system to process that whole query and provides desired images to users and also not user friendly. Hence, we have proposed a new technique for image searching where system accepts multimodal input that can be voice, text and image. This system is useful in a case where users do not know exact name of an image but by describing it using either text or speech or by providing any other relevant image, users can easily find targeted image. The ANN technique is also added into it to increase the performance of the system. The system works in three phases- 1) Image Composition, 2) Image Processing, 3) ANN. Lastly, before providing images to users as an output, images are divided into two sets- positive and negative. All relevant images are present in positive set and non relevant images are present in negative set. Lastly images present in positive set are provided to the user as a final output. Hence our approach improves the quality of image searching technique and provides easy way to finds targeted image.
Keywords: Image Processing, Artificial Intelligence, Content Based Image Retrieval, Multimodal Input