AdaptiveEqualization may fail on N0f8 images

[vask827928]

Using ImageContrastAdjustment v0.3.12 with JuliaImages.jl and I am having some troubles using Adaptive histogram Equalization:

I1 = load("/some-image.jpg"); 
I1=Gray.(I1);
alg = AdaptiveEqualization(nbins = 256, rblocks = 8, cblocks = 8, clip = 0.2)
I1_eq = adjust_histogram((I1), alg)

this doesnt happen always, but for some images I get ArgumentError:

ArgumentError: N0f8 is an 8-bit type representing 256 values from 0.0 to 1.0; cannot represent 1.0000000000000002

Stacktrace:
  [1] throw_converterror(::Type{N0f8}, x::Float64)
    @ FixedPointNumbers ~/.julia/packages/FixedPointNumbers/Dn4hv/src/FixedPointNumbers.jl:326
  [2] _convert
    @ ~/.julia/packages/FixedPointNumbers/Dn4hv/src/normed.jl:76 [inlined]
  [3] FixedPoint
    @ ~/.julia/packages/FixedPointNumbers/Dn4hv/src/FixedPointNumbers.jl:58 [inlined]
  [4] convert
    @ ./number.jl:7 [inlined]
  [5] Gray
    @ ~/.julia/packages/ColorTypes/L7biZ/src/types.jl:359 [inlined]
  [6] convert
    @ ~/.julia/packages/ColorTypes/L7biZ/src/conversions.jl:105 [inlined]
  [7] setindex!
    @ ./array.jl:996 [inlined]
  [8] transform_interior!(out::Matrix{…}, img::Matrix{…}, bounds::Tuple{…}, block_centroids::Tuple{…}, block_dimensions::Tuple{…}, intensity_range::Tuple{…}, block_cdf::Matrix{…})
    @ ImageContrastAdjustment ~/.julia/packages/ImageContrastAdjustment/J15Ip/src/algorithms/adaptive_equalization.jl:515
  [9] transform_image!(out::Matrix{…}, img::Matrix{…}, block_centroid_r::Vector{…}, block_centroid_c::Vector{…}, block_width::Int64, block_height::Int64, intensity_range::Tuple{…}, cblocks::Int64, rblocks::Int64, block_cdf::Matrix{…})
    @ ImageContrastAdjustment ~/.julia/packages/ImageContrastAdjustment/J15Ip/src/algorithms/adaptive_equalization.jl:449
 [10] (::AdaptiveEqualization{Float64, Float64, Float64})(out::Matrix{Gray{N0f8}}, img::Matrix{Gray{N0f8}})
    @ ImageContrastAdjustment ~/.julia/packages/ImageContrastAdjustment/J15Ip/src/algorithms/adaptive_equalization.jl:299
 [11] adjust_histogram!
    @ ~/.julia/packages/ImageContrastAdjustment/J15Ip/src/HistogramAdjustmentAPI/histogram_adjustment.jl:44 [inlined]
 [12] #adjust_histogram#3
    @ ~/.julia/packages/ImageContrastAdjustment/J15Ip/src/HistogramAdjustmentAPI/histogram_adjustment.jl:64 [inlined]
 [13] adjust_histogram
    @ ~/.julia/packages/ImageContrastAdjustment/J15Ip/src/HistogramAdjustmentAPI/histogram_adjustment.jl:59 [inlined]
 [14] adjust_histogram(::Matrix{Gray{N0f8}}, ::AdaptiveEqualization{Float64, Float64, Float64})
    @ ImageContrastAdjustment.HistogramAdjustmentAPI ~/.julia/packages/ImageContrastAdjustment/J15Ip/src/HistogramAdjustmentAPI/histogram_adjustment.jl:68
 [15] top-level scope
    @ In[52]:2
 [16] eval(m::Module, e::Any)
    @ Core ./boot.jl:489

is this a bug, or by design? if i change the type from N0f8 to Float it works correctly.

[zygmuntszpak]

Thanks for reporting the issue. Would it be possible to share an example image for which it fails?

[vask827928]

I was trying with a set of these three images:

Attempting AdaptiveEqualization(nbins = 256, rblocks = 8, cblocks = 8, clip = 0.2) on all but the second image fail (unless I change the numeric type).

I also tried out changing parameters like AdaptiveEqualization(nbins = 256, rblocks = 4, cblocks = 4, clip = 0.2). The latter only fails on the first image, but the second and third work.