New fast image edge-detection algorithm based on composite self-adaption predictor

摘要

Using advantages of gradient adjusted predictor (GAP) and gradient edge detection (GED) predictors of lossless image coding for reference, and the image was cut into four equal parts with the application of Graphics Processor Unit (GPU) parallel technology operation. In four sub-images, the composite self-adaption predictor was employed for predicting error image, threshold classification error image edge and thinning edge. Results showed that with the application of the parallel technology which avoided errors multiply, not only the complexity of the time was reduced significantly, but also the distinct, holistic and detail-rich edge image was obtained.