A method including receiving a first two-dimensional (2D) image; and applying a filter to the 2D image to produce a filtered image that identifies a circular object of interest, wherein the filter is based on the integral sum of the function S, where the filter output at point x is M ⁡ ( x ) = ∫ ∀ y ∈ V ′ ⁢ S ⁡ ( m , θ , r , y ) ⁢ Δ ⁢ ⁢ y which is obtained from the 2D image, the function S is represented by S(m,θ,r,y)=S1(m)S2(θ,r), where m is a magnitude of a gradient at location y, r is a radial distance from y to x, and θ is an angle between the gradient at location y and the radial distance from y to x, S 1 ⁡ ( m ) = ( tan - 1 ⁡ ( m - C 1 ) + π 2 ) π , ⁢ s 2 ⁡ ( θ , r ) = 1 σ ⁢ 2 ⁢ ⁢ π ⁢ ⅇ - ( r × sin ⁢ ⁢ θ ) 2 2 ⁢ ⁢ σ 2 × ∫ - r × sin ⁡ ( 90 - θ ) C 2 - r × sin ⁡ ( 90 - θ ) ⁢ 1 σ ⁢ 2 ⁢ ⁢ π ⁢ ⅇ - x 2 2 ⁢ ⁢ σ 2 ⁢ ⁢ ⅆ x , C1 depends on a gray-level value of the object, C2 and σ depend on a size of the object.