REDUCED HARDWARE ARCHITECTURE OF BILATERAL FILTER FOR REAL TIME IMAGE DENOISING

A detailed description of a bilateral filter for grayscale and color image denoising is stated and this helps to denoise the image while preserving its edges. The procedure is non-repetitive, local, simple and reduces hardware utilization. Based on both geometric proximity and photometric analogy, bilateral filter amalgamates gray levels and favor adjacent values to far-off values in both domain and range. The distinctive feature of our design concept consists of processing the complete filter window in one pixel clock cycle. This characteristic of the kernel-based scheme is hold up by the positioning the input data into groups. Additionally, by the exploitation of distinguishability and evenness of filter component, the intricacy of the design is extensively diminished. Integrating these properties, the bilateral filter is executed with very inexpensive and valuable usage of dedicated resources. Due to the balance of the filter design, kernels of various dimensions can be accomplished with less effort using our design. The final denoised image properties depends on the selected filter specifications only.