Iterative content adaptable purple fringe detection

Correction of complex purple fringing by green‐channel compensation and local luminance adaptation

“…This makes the information separation problem challenging. Absence of a clear and PFA‐free reference unavailable : Since the original natural scene is unavailable, this process of un‐mixing and separation becomes truly blind in nature and the entire parameter estimation and equalisation procedure becomes INTRINSIC to the image under test. On account of the mixing of scene information with a purple semi‐transparent cloak, conventional PFA detection techniques which are manual‐threshold dependent [13–22] cannot be used for picking up the complete region. Unless the detection procedure is completely content adaptive and scene/content independent [12], this complete C‐PFA region cannot be picked up. Thus, this leads to a partial restoration of the affected image. The literature review is split into two segments, one part covering the correction solutions for the simpler IS‐PFA problem and secondly, selective patented work linked to the purple haze (C‐PFA) correction problem.…”

“…It is imperative that the correction algorithm does not alter parts which are free from PFA. There are two ways in which this issue can be handled: Using a content adaptive PFA detection algorithm [11, 12], for isolating and segregating the parts which have been affected by C‐PFA: This should work despite having a complex texture in the backdrop. If the induction of PFA pixels is overly aggressive, then there are likely to be false positives, wherein regular pixels are picked up as PFA pixels and modified un‐necessarily.…”

“…If the RULE is satisfied the pixel colour is modified, else it is kept as it is: this line of thinking has been adopted by Athentech [23], which examines the green‐channel value in relation to Red and Blue channels at the same location and forms an opinion regarding the correction enforcement on the Red and Blue channels. In this work, the first route has been adopted, wherein a content adaptive PFA detection algorithm has been deployed for picking up C‐PFA fringes despite the background scene variation in the backdrop. It was shown in [12], that this adaptive detection algorithm works well even for images of a complex type and the key parameter in question is the chromatic distance ratio threshold, which can be adjusted to incorporate fringes of a diverse type. To maintain the right balance by minimising false positives, while maximising the girth of the PFA detection algorithm (to capture the complete fringe affected region, including even the faintest form of PFA), the chromatic distance ratio threshold should not exceed 0.95 [12].…”

“…It was shown in [12], that this adaptive detection algorithm works well even for images of a complex type and the key parameter in question is the chromatic distance ratio threshold, which can be adjusted to incorporate fringes of a diverse type. To maintain the right balance by minimising false positives, while maximising the girth of the PFA detection algorithm (to capture the complete fringe affected region, including even the faintest form of PFA), the chromatic distance ratio threshold should not exceed 0.95 [12].…”

“…4 d , 5 d and 6 d ). For both the types of PFA, the detection algorithm used the one based on chromatic distance ratios (CDRs) discussed in brief in the later sections and in detail in Malik and Karthik [12]. While there exist a gamut of solutions to correct IS‐PFA, there are none geared towards blind quality assessment and correction of the more complex form of PFA, viz.…”