| 摘要 |
PURPOSE: To make an image processor possible to provide an image which is accurate and large in transmitted information amount by removing a noise or peculiarity characteristic of the device system from image information obtained by irradiating a body with an information carrier or transmitting the carrier through the body and measuring its variation, reducing the noise without causing blurring due to a shake, and converting the image into an image which well matches user's taste. CONSTITUTION: An image conversion processing part 2 which inputs image data from an original image supply part 1 and performs image conversion processing is composed of a neural network 4 which couples artificial neural elements in a layer structure of an input layer, an intermediate layer, and an output layer, and inputs time-series data of a specific area including an at least noticed pixel from the original image supply part 1 to perform the image conversion processing, and outputs data corresponding to the noticed pixel. Consequently, an original image is converted through the image conversion processing of the neural network 4 to provide the image which is accurate and large in transmitted information quantity. |
