System and Method for Image Denoising Optimizing Object Curvature

A method for removing noise from an image, comprising:

• receiving image data including a plurality of pixels;
• identifying a plurality of neighborhoods of three neighboring pixels of the received image data;
• reducing an intensity value of a center pixel of a neighborhood of three neighboring pixels if and only if the value of the center pixel is greater than the values of both of the other pixels of the neighborhood of three neighboring pixels;
• increasing the intensity value of the center pixel of the neighborhood of three neighboring pixels if and only if the value of the center pixel is less than the values of both of the other pixels of the neighborhood of three neighboring pixels;
• and
• neither reducing nor increasing the intensity value of the center pixel of the neighborhood of three neighboring pixels when the value of the center pixel is between the values of the other pixels of the neighborhood of three neighboring pixels or the value of the center pixel of the neighborhood of three neighboring pixels is equal to the values of one or both of the other pixels of the neighborhood of three neighboring pixels.

A computer system comprising:

• a processor;
• and
• a non-transitory, tangible, program storage medium, readable by the computer system, embodying a program of instructions executable by the processor to perform method steps for removing noise from an image, the method comprising:
• receiving image including a plurality of pixels;
• determining a penalty for each neighborhood of three neighboring pixel within the received image, the neighborhood including a middle pixel and two neighboring pixels, the penalty being zero when a value of the middle pixel is between or equal to the values of the neighboring pixels and the penalty being greater than zero when the value of the middle pixel is not between or equal to the values of the neighboring pixels;
• minimizing the determined penalty over the entire received image;
• and
• providing a denoised image based on the penalty-minimized image.