Based on CNN Feature Learning, a Local Metric for Defocus Blur Detection PROJECT TITLE : A Local Metric for Defocus Blur Detection Based on CNN Feature Learning ABSTRACT: Blur detection in computer vision and digital imaging areas is a complex and demanding problem. The creation of local sharpness metric maps has been a major focus of previous work on defocus blur detection. Using multiple convolutional neural networks, this research proposes a simple yet effective way to automatically obtain the local metric map for defocus blur detection (ConvNets). In a supervised manner, ConvNets learn the most important features at the super-pixel level of a picture. We can automatically derive the local sharpness measure by altering the principal component vector by extracting convolution kernels from the trained neural network structures and using principal component analysis. It is also proposed to use the intrinsic properties of the hyperbolic tangent function to refine the defocus blur detection result from coarse to fine. Tests show that compared to earlier methods, our proposed solution consistently outperforms them. Did you like this research project? To get this research project Guidelines, Training and Code... Click Here facebook twitter google+ linkedin stumble pinterest A Morphological Reconstruction-Based Image Dehazing Algorithm Based on Epipolar Plane Images, a Maximum Likelihood Approach for Depth Field Estimation