Image Decomposition Sensor Based on Conditional Adversarial Model

In this paper, we propose an Image Decomposition Sensor based on Conditional Adversarial Model (CAM). The presented device decomposes intrinsic image into reflectance layer and shadow layer and the Conditional Adversarial Model (CAM) trains the intrinsic image decomposition to obtain high quality reflectance image and shadow image. By enhancing the shadow image, we can get a larger range of contrast images. Besides, we adopt reduced lookup table based pipeline architecture to greatly reduce the overall computing time. The image decomposition device solves the problems such as limited contrast and color distortion and is used for application of image contrast enhancement. The sensor circuit provide excellent auxiliary effects for system that need to rely on high-speed image processing.